False
        

Triggered text Ignored text Blocked text

tag:github.com,2008:https://github.com/ggml-org/whisper.cpp/releases

Release notes from whisper.cpp

2026-09-11T05:31:53Z tag:github.com,2008:Repository/541269386/v1.9.4 2026-09-11T05:31:55Z

v1.9.4

<h2>Overview</h2> <p>New version has been released.</p> <p><strong>Nightly build:</strong> <a href="https://github.com/ggml-org/whisper.cpp/releases/tag/b5130">b5130</a><br> <strong>More info:</strong> <a href="https://github.com/ggml-org/ggml/discussions/1579">dist : releases and versioning of ggml-org projects</a></p> <h2>Changelog since v1.9.3</h2> <p><a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/927cfce34f31707e17f2bff35c349632fb9e2c3a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/927cfce34f31707e17f2bff35c349632fb9e2c3a"><tt>927cfce</tt></a> metal : remove leftover ggml-metal.metal kernels file (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5412307245" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4051" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4051/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4051">#4051</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/dd80eb2afd6d1a02ac5d4eb764e757c2425d2236/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/dd80eb2afd6d1a02ac5d4eb764e757c2425d2236"><tt>dd80eb2</tt></a> scripts : use sed instead of grep for version parsing [no ci] (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5412397353" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4052" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4052/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4052">#4052</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1fa6dfa05f3519910485d2054172d806c5c6ac4f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1fa6dfa05f3519910485d2054172d806c5c6ac4f"><tt>1fa6dfa</tt></a> ci : update WoA CUDA 13.4 to use 13.4.1 GA (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5412496064" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4053" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4053/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4053">#4053</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a2b36eb677918d4f9ab1db7b8a7ff968563ed163/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a2b36eb677918d4f9ab1db7b8a7ff968563ed163"><tt>a2b36eb</tt></a> whisper : bump version to 1.9.4 (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5409539635" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4050" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4050/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4050">#4050</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/6fb4cd675c1f4d22c2b3217665a61b01faed34db/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/6fb4cd675c1f4d22c2b3217665a61b01faed34db"><tt>6fb4cd6</tt></a> ci : add Windows On ARM support to release job (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5395574377" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4048" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4048/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4048">#4048</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c44b60b8053bbf2a5c1e014f11323fb3f2485177/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c44b60b8053bbf2a5c1e014f11323fb3f2485177"><tt>c44b60b</tt></a> whisper : call encoder_begin_callback before language auto-detect (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4838874999" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3936" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3936/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3936">#3936</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/6d0ed914991cac42ace012a05588880801831343/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/6d0ed914991cac42ace012a05588880801831343"><tt>6d0ed91</tt></a> whisper : re-seed decoder 0 between calls (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5288423327" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4025" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4025/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4025">#4025</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/cec4dbe5d5a18b0c1b15a7e9c526c35bd8d46a96/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/cec4dbe5d5a18b0c1b15a7e9c526c35bd8d46a96"><tt>cec4dbe</tt></a> ci : update close-issue job to not close issues (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5382767932" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4045" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4045/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4045">#4045</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/79f2d921129284a850f654c6ff80ba6cfdeda16f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/79f2d921129284a850f654c6ff80ba6cfdeda16f"><tt>79f2d92</tt></a> tests : load backends before init when built with GGML_BACKEND_DL (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5331432738" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4031" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4031/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4031">#4031</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/61e6ccade0849f0f518c6decd1f25b3f9fb77b76/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/61e6ccade0849f0f518c6decd1f25b3f9fb77b76"><tt>61e6cca</tt></a> server : return language in detect response (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5354191995" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4035" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4035/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4035">#4035</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/52a939a2a762224e255d366c1182b2af4dd1a032/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/52a939a2a762224e255d366c1182b2af4dd1a032"><tt>52a939a</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a937f4e8efba28c0da4187617434676572355107/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a937f4e8efba28c0da4187617434676572355107"><tt>a937f4e</tt></a> ggml : bump version to 0.23.0 (ggml/1618)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/11d4eec830119e58f880df52df97f63a3a26f6c5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/11d4eec830119e58f880df52df97f63a3a26f6c5"><tt>11d4eec</tt></a> metal : add remaining fa-vec tunings for M3 Max (llama/28373)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/140e57a4e20af04e8e00930798474a4707f0ee47/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/140e57a4e20af04e8e00930798474a4707f0ee47"><tt>140e57a</tt></a> ggml : replace compile definitions with version.h.in (llama/28364)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e2389eb99cf5796659495c0066bb491fb52a354f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e2389eb99cf5796659495c0066bb491fb52a354f"><tt>e2389eb</tt></a> ggml : rename and make private ggml_op_alloc_size_may_expand() (ggml/0)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1b37beadd76496ed4081998150c2b8a81783b7a0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1b37beadd76496ed4081998150c2b8a81783b7a0"><tt>1b37bea</tt></a> ggml : don't crash when backend search path can't be read (llama/28271)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f32e6fa0c8f1a78fafe4a911821e660b76a2c02d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f32e6fa0c8f1a78fafe4a911821e660b76a2c02d"><tt>f32e6fa</tt></a> ggml : remove GGML_CUDA_PEER_MAX_BATCH_SIZE (llama/28177)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e1bbe405209c6e6836f9416a5df4363fe44cddb0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e1bbe405209c6e6836f9416a5df4363fe44cddb0"><tt>e1bbe40</tt></a> ggml-cpu(s390x) : fix q5_1 uninitialized v_acc (llama/28332)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d1e0e6491ff3735a09db4b160524e0dd4488a4e4/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d1e0e6491ff3735a09db4b160524e0dd4488a4e4"><tt>d1e0e64</tt></a> sycl: fuse rms_norm+mul+add and add+add residual chains (llama/27610)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/36f170e5fe39b566eb2afe1d2ed74b567cd54fc9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/36f170e5fe39b566eb2afe1d2ed74b567cd54fc9"><tt>36f170e</tt></a> SYCL: Refactor GGML_SYCL_ENABLE_MKL_FA to global var (llama/26863)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d784add75f41c7487da2b2b5c9ebc224f3b4e289/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d784add75f41c7487da2b2b5c9ebc224f3b4e289"><tt>d784add</tt></a> opencl: quant lm_head / decode GEMV and medium-batch GEMM optimizations (speculative decoding/MTP) (llama/26477)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0a4a95c86e592a8d2136e3316d2d6496c5033d14/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0a4a95c86e592a8d2136e3316d2d6496c5033d14"><tt>0a4a95c</tt></a> tune MMVQ to MMQ crossover for SM87 (llama/28285)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4dd48dde357afef5c208adad869082edf94e70b1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4dd48dde357afef5c208adad869082edf94e70b1"><tt>4dd48dd</tt></a> metal : add sparse FA (llama/28098)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d55d345e696debc6c8be4899b60c0238e621c792/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d55d345e696debc6c8be4899b60c0238e621c792"><tt>d55d345</tt></a> metal : fix glu dispatch with ne00 = 1 (llama/28306)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/25350b579e7e957b4f005400fe72b40162082b18/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/25350b579e7e957b4f005400fe72b40162082b18"><tt>25350b5</tt></a> CUDA: Allow concurrent streams per split for multi-GPU (llama/28198)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/47d348a2780fd70c8b50906a5c12cf0cc9a1b3fa/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/47d348a2780fd70c8b50906a5c12cf0cc9a1b3fa"><tt>47d348a</tt></a> vulkan: fix FA dequant path engagement (llama/28190)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f24a38605b729b4ab6546ccbde2d9f92d773558f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f24a38605b729b4ab6546ccbde2d9f92d773558f"><tt>f24a386</tt></a> sycl : enhance the api to support peer-to-peer copy (llama/27550)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a704770e3bdcce27c61b2a7695c6885ee79b6432/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a704770e3bdcce27c61b2a7695c6885ee79b6432"><tt>a704770</tt></a> sycl: reduce redundant work in Q4_K multi-column MMVQ (llama/27062)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e5605697ce4dce019370afae61b7d222f2fb4bbf/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e5605697ce4dce019370afae61b7d222f2fb4bbf"><tt>e560569</tt></a> finetune: fix no KV cache (llama/27199)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/37f0f443d72003a83f1d95a8681ab194a254d7d6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/37f0f443d72003a83f1d95a8681ab194a254d7d6"><tt>37f0f44</tt></a> ggml-hexagon: add F16 support for unary ops (llama/28228)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1bdda1e36676acd6f18d004494ab31847a92ae00/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1bdda1e36676acd6f18d004494ab31847a92ae00"><tt>1bdda1e</tt></a> metal : add fa-vec tunings for M3 (llama/28236)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3a1c7d6b6f47efeb69bde410b0003129563fa9f7/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3a1c7d6b6f47efeb69bde410b0003129563fa9f7"><tt>3a1c7d6</tt></a> metal : fix memory query under low-memory conditions (llama/27701)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4d343d7c05c36bc863536cab9d042b96acf0eb9c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4d343d7c05c36bc863536cab9d042b96acf0eb9c"><tt>4d343d7</tt></a> ggml-cuda : remove unused vars (llama/28235)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/519df618df430b9288cb62de6eb4e5568965526c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/519df618df430b9288cb62de6eb4e5568965526c"><tt>519df61</tt></a> CUDA + ggml: add sparse-fa for DSV4/GLM (llama/27970)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c2b400754bac2de1f4f74bfac9eb33474daa631e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c2b400754bac2de1f4f74bfac9eb33474daa631e"><tt>c2b4007</tt></a> ggml: avoid KleidiAI buffer type init on dispatch (llama/27891)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/dc70853ec28bdfc716641111f37cee0fe4bd8a91/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/dc70853ec28bdfc716641111f37cee0fe4bd8a91"><tt>dc70853</tt></a> hexagon: MUL_MAT and MUL_MAT_ID fusion and fixes (llama/28202)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1c7d35e14af94778e40f92afb727fd9eb159783b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1c7d35e14af94778e40f92afb727fd9eb159783b"><tt>1c7d35e</tt></a> vulkan: handle larger batch sizes (>4) efficiently for IQ3_S mat-vec (llama/27449)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a9e58612c5145ba47b101889b3f08386a7f42f23/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a9e58612c5145ba47b101889b3f08386a7f42f23"><tt>a9e5861</tt></a> vulkan : only request VK_KHR_shader_bfloat16 extension if supported (llama/28155)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d57ae98246c503fad649a8d715ff995b1b867ba6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d57ae98246c503fad649a8d715ff995b1b867ba6"><tt>d57ae98</tt></a> ggml-cpu : conditionally add SpacemiT IME kernel sources (llama/27961)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/35133c94c5aa00f39ad4d79513c5e57cd6de0cb3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/35133c94c5aa00f39ad4d79513c5e57cd6de0cb3"><tt>35133c9</tt></a> opencl: fix out‐of‐bound reads in the Adreno image kernels (#27632)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c94921f8c6787ee12af882672ed91af9982b91cc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c94921f8c6787ee12af882672ed91af9982b91cc"><tt>c94921f</tt></a> hexagon: add missing FARF logs for cpy/get_rows/set_rows/gdn ops (llama/28217)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fcc2feee2fee5e6df76847aeaaa5bdb1eead3352/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fcc2feee2fee5e6df76847aeaaa5bdb1eead3352"><tt>fcc2fee</tt></a> metal : add metallib build support for xcframework (llama/28163)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2c486783abfafacdffbc661f0bba0b31e8994eb7/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2c486783abfafacdffbc661f0bba0b31e8994eb7"><tt>2c48678</tt></a> cuda: fuse MoE weighted expert reduction (llama/25952)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f162a194754838c2468d1d02600182154ba41a63/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f162a194754838c2468d1d02600182154ba41a63"><tt>f162a19</tt></a> Revert "sycl : add Kronecker product FWHT support for sizes 384, 640, 768, 12…" (#28184)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/408faaafcedc643e8feb77a3b2536a855fcd6549/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/408faaafcedc643e8feb77a3b2536a855fcd6549"><tt>408faaa</tt></a> sycl : add Kronecker product FWHT support for sizes 384, 640, 768, 1280 (llama/28016)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5f07f856c50c0480af1b0fa883f84068a1cd0e7f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5f07f856c50c0480af1b0fa883f84068a1cd0e7f"><tt>5f07f85</tt></a> metal : add fa-vec tuning for M2 Pro (llama/28122)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8cca1a3616d854200a48922966552daf6cff2433/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8cca1a3616d854200a48922966552daf6cff2433"><tt>8cca1a3</tt></a> metal : add fa-vec tunings for A18 Pro (MacBook Neo) (llama/28152)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a245a8f48f4c59bb8da2eaa710d053c726f37c59/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a245a8f48f4c59bb8da2eaa710d053c726f37c59"><tt>a245a8f</tt></a> metal : fix more leaks due to missing autoreleasepools (llama/27883)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/870db2afa1a03001b941b920abb761729ba69f9a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/870db2afa1a03001b941b920abb761729ba69f9a"><tt>870db2a</tt></a> metal : add fa-vec tuning for M2 Max (llama/28015)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4f3a2a4b79e32efae8d291cae9322689454be368/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4f3a2a4b79e32efae8d291cae9322689454be368"><tt>4f3a2a4</tt></a> sycl : support limit max alloc memory within 2GB for host-pinned memory (llama/27559)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5032008bc86f37d62f2fc3a4cbab6131e5dc10c9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5032008bc86f37d62f2fc3a4cbab6131e5dc10c9"><tt>5032008</tt></a> metal: enable Metal 4.0 tensor API on M5+/A19+ (llama/27461)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8e54c659b583c1a159f7164f9de338844f30f779/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8e54c659b583c1a159f7164f9de338844f30f779"><tt>8e54c65</tt></a> metal : add fa-vec tunings for M1 Ultra (llama/28088)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f22bb2ea4d96da2bf8cce738805c96c0c805a203/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f22bb2ea4d96da2bf8cce738805c96c0c805a203"><tt>f22bb2e</tt></a> CUDA: XOR swizzle flash attn K,V smem fp16 tiles (llama/25635)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/dbc40efce86f37fad93ecf211d3064f8ac79cff6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/dbc40efce86f37fad93ecf211d3064f8ac79cff6"><tt>dbc40ef</tt></a> metal : add concat support for quantized types (llama/28116)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2f608ab4f8a1107f87aa372c7f3d7492e63438fe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2f608ab4f8a1107f87aa372c7f3d7492e63438fe"><tt>2f608ab</tt></a> AVX2: Speed up large batch size prompt processing of IQ models (llama/27402)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c6934d0fcf562ada5493208f6aa5804947f37b64/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c6934d0fcf562ada5493208f6aa5804947f37b64"><tt>c6934d0</tt></a> metal : add top-k radix implementation (llama/28073)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/088c603e2951f3bf5b7ce59438523c541580faa1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/088c603e2951f3bf5b7ce59438523c541580faa1"><tt>088c603</tt></a> opencl: tune the quant paths for Intel Xe-LP GPUs to improve its TG and PP performance (llama/26438)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c648b9a4d0fbb0dd3db9d19149714f77cef43af2/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c648b9a4d0fbb0dd3db9d19149714f77cef43af2"><tt>c648b9a</tt></a> webgpu : avoid crash when offset is not multiple of 4 in WebGPU ggml_backend_tensor_get() implementation (llama/28045)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c1be45b89016bb11b1460618c97b1e077adcd9a7/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c1be45b89016bb11b1460618c97b1e077adcd9a7"><tt>c1be45b</tt></a> ROCm: add radix TOP_K for long rows (llama/27466)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/7614a4c1398708ea1a6575d0e3b659f14bb2f7e3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/7614a4c1398708ea1a6575d0e3b659f14bb2f7e3"><tt>7614a4c</tt></a> metal : add fa-vec tunings for M1 (llama/28078)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b0f4bc02ed1f15a0d63eca3cc8af7e4d6a481d59/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b0f4bc02ed1f15a0d63eca3cc8af7e4d6a481d59"><tt>b0f4bc0</tt></a> CUDA: extend MOE fusion to specdec, earlier MOE glu fusion and topk-router fusion were restricted to 1 token (llama/27621)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/76a51e82d7cb548eecc880e108fdfc0d7040be57/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/76a51e82d7cb548eecc880e108fdfc0d7040be57"><tt>76a51e8</tt></a> sycl : Enhance to get the free memory of Intel GPU (llama/27968)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/6ce7b89952fd07d0db62e99fb15c231ca3979714/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/6ce7b89952fd07d0db62e99fb15c231ca3979714"><tt>6ce7b89</tt></a> vulkan: tune mat-vec rows for batched inference on Strix Halo (llama/27909)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/96dddd87f20b470a721a3c2bd25aba63a045a6a8/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/96dddd87f20b470a721a3c2bd25aba63a045a6a8"><tt>96dddd8</tt></a> ggml : add MUL_MAT to the list of ops that may need additional memory (for WebGPU) (llama/28071)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/db00b0196b1be6af1bf184c8948a50fc3120fcb5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/db00b0196b1be6af1bf184c8948a50fc3120fcb5"><tt>db00b01</tt></a> vulkan: top_k radix select for k >= 1024 for Qwen 3.8 Flash Next (llama/28032)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/01ebd225a7784bfb8245b98f2ebd8a4589f32421/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/01ebd225a7784bfb8245b98f2ebd8a4589f32421"><tt>01ebd22</tt></a> hexagon: fix CPY fence bug (llama/28033)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e5c96ca45da0d38b0d4a0967e83ec413e530b3d9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e5c96ca45da0d38b0d4a0967e83ec413e530b3d9"><tt>e5c96ca</tt></a> metal : add remaining Q4_1/Q5_0/Q5_1 fa-vec tunings for M2 (llama/28017)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4089fa628a00d026b8090616fc4d595ab36ccbd1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4089fa628a00d026b8090616fc4d595ab36ccbd1"><tt>4089fa6</tt></a> rpc: avoid serializing buffers from other servers (llama/26500)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/749683d30f40845d14f0da8d7df64f219c6db1c5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/749683d30f40845d14f0da8d7df64f219c6db1c5"><tt>749683d</tt></a> ggml : fix ggml_backend_buft_get_alloc_size() guard (llama/28038)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e9583f075aafaea362afd80a87fdac74a0f54256/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e9583f075aafaea362afd80a87fdac74a0f54256"><tt>e9583f0</tt></a> ggml: add SWIGLU_CLAMP (llama/27930)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e900a732c8908510bc4e932544daab86c8790a12/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e900a732c8908510bc4e932544daab86c8790a12"><tt>e900a73</tt></a> CUDA: use the fast mm_ids_helper path for any n_expert_used (llama/27978)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/35d9e2237e7e8a6a716cd3d517ba46ba62bb1bed/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/35d9e2237e7e8a6a716cd3d517ba46ba62bb1bed"><tt>35d9e22</tt></a> hip: tune rdna 3 mmq config (llama/26284)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e5c9e3e3e9a9c3f1763c54803229ce393f3669d1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e5c9e3e3e9a9c3f1763c54803229ce393f3669d1"><tt>e5c9e3e</tt></a> hip : optimize Q2_0 dot-product path for gfx1201 (llama/26753)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4b2243a6c20c4503f42ab47eeeddc274f5a917b1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4b2243a6c20c4503f42ab47eeeddc274f5a917b1"><tt>4b2243a</tt></a> ggml : add ggml_backend_op_alloc_size_may_expand, use it in RPC (llama/27960)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/43acf3d6e84b210e72a5263f72f4d8fa97e540fd/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/43acf3d6e84b210e72a5263f72f4d8fa97e540fd"><tt>43acf3d</tt></a> rpc: fix apple rdma error spew on teardown (llama/27908)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1e0f38257378496e822093d61fb154f2e55c7236/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1e0f38257378496e822093d61fb154f2e55c7236"><tt>1e0f382</tt></a> metal: add fa-vec tunings for M3 Ultra (llama/27999)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b66593ef1bd6971f3eafa3ab79f0c86c92a4e8e0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b66593ef1bd6971f3eafa3ab79f0c86c92a4e8e0"><tt>b66593e</tt></a> metal : Add fa-vec tuning for M3 Pro (llama/27963)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5e494599a4521878eb937765b5811f28ff314888/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5e494599a4521878eb937765b5811f28ff314888"><tt>5e49459</tt></a> rpc : fix pre-rdma macOS versions (llama/27815)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3ad8b9b217e7d0d8a829e9344dbe35bfdf19dc64/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3ad8b9b217e7d0d8a829e9344dbe35bfdf19dc64"><tt>3ad8b9b</tt></a> hexagon: support for device discovery and create sessions on demand (llama/27785)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3d4e0e9858ffc1d4446d8b4ce7d4385d696e3cb6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3d4e0e9858ffc1d4446d8b4ce7d4385d696e3cb6"><tt>3d4e0e9</tt></a> sycl: split long rows in TOP_K instead of one work-group per row (llama/27847)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c68f20555aa9ff91dbec5d6c23e8a862174e4d8f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c68f20555aa9ff91dbec5d6c23e8a862174e4d8f"><tt>c68f205</tt></a> metal : fix null-pipeline crash for F16 src1 mul_mat/mul_mat_id (llama/25648)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c969c68b544cd2e90052b58647b52ce67bd93575/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c969c68b544cd2e90052b58647b52ce67bd93575"><tt>c969c68</tt></a> ggml: allow passing alloc dependencies in graph_optimize (llama/27301)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b33bbc5c4eef7a435de6f907d6ff4547274684ed/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b33bbc5c4eef7a435de6f907d6ff4547274684ed"><tt>b33bbc5</tt></a> metal : add fa-vec tunings for M2 (llama/27940)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2a11026cfe3a660673e8c8c37dac18541ea3b1f6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2a11026cfe3a660673e8c8c37dac18541ea3b1f6"><tt>2a11026</tt></a> opencl: use a better matmul path on two Adreno GPU generations (llama/27640)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/285f1ffd99d6b1ce97a59fc5978574140c0f7846/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/285f1ffd99d6b1ce97a59fc5978574140c0f7846"><tt>285f1ff</tt></a> metal : assert shared memory padding (llama/27951)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/308fa4f8a2beec1f76062727d4b1ca0a4dcbafff/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/308fa4f8a2beec1f76062727d4b1ca0a4dcbafff"><tt>308fa4f</tt></a> metal : add remaining fa-vec tunings for M4 Pro (llama/27915)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/325c8d16c1264ae0b0f057b45bcb87be32a8f89f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/325c8d16c1264ae0b0f057b45bcb87be32a8f89f"><tt>325c8d1</tt></a> sycl: make --fit respect --fit-target better (llama/27629)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e644752070ed07294219e787cc10d2697888f96d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e644752070ed07294219e787cc10d2697888f96d"><tt>e644752</tt></a> vulkan: combine duplicated fastdiv functions, rename the one optimizing small divs (llama/27526)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/590fe189021127f8a8255968e124c892293a9703/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/590fe189021127f8a8255968e124c892293a9703"><tt>590fe18</tt></a> metal : add fa-vec tunings for M1 Max (llama/27932)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d501a0a3e8a50e5fcef657e558ff95f51a91c4d5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d501a0a3e8a50e5fcef657e558ff95f51a91c4d5"><tt>d501a0a</tt></a> vulkan: Change mul_mat_id to pad K rather than N (llama/27925)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4c38040fd3b3241a171bbc9ef218cf34dfdd5994/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4c38040fd3b3241a171bbc9ef218cf34dfdd5994"><tt>4c38040</tt></a> vulkan: fix missing view-alias dependencies in ggml_vk_graph_optimize (llama/27812)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/caea96f6ceef38c2fe6d6e0332c83ae42436b264/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/caea96f6ceef38c2fe6d6e0332c83ae42436b264"><tt>caea96f</tt></a> ggml : fix conv_transpose_2d for multiple batches (llama/26132)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ba99c098669484ad18978f2b18557726709fba12/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ba99c098669484ad18978f2b18557726709fba12"><tt>ba99c09</tt></a> Vulkan: add hoisting support for row IDs and expert count in shaders (llama/26686)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0a150873efb17f3b21e92c83d705bbfdce200bd0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0a150873efb17f3b21e92c83d705bbfdce200bd0"><tt>0a15087</tt></a> metal : add fa-vec tunings for M4 (llama/27875)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/7f78e1b4694bdae54e8c9101b96be14f89b76f02/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/7f78e1b4694bdae54e8c9101b96be14f89b76f02"><tt>7f78e1b</tt></a> OpenVINO: Update OV to 2026.3.1, whisper.cpp support, Qwen3.5 on NPU, and new ops (llama/27843)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fa4d244c939ab9f14e0ce31a08ca85a5bb16353b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fa4d244c939ab9f14e0ce31a08ca85a5bb16353b"><tt>fa4d244</tt></a> sycl: use TILE for quantized KV decode on BMG (llama/26689)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/97d0da26a2e90861a7616a2e12ca4f1e2ed3cbde/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/97d0da26a2e90861a7616a2e12ca4f1e2ed3cbde"><tt>97d0da2</tt></a> sycl: bind the f16 KV cache in place for the oneDNN SDPA path (llama/27468)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/530e3f4834477a267889796e3edd3bdde228ed75/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/530e3f4834477a267889796e3edd3bdde228ed75"><tt>530e3f4</tt></a> metal : add fa-vec tunings for M3 Max, M5 and M5 Pro (llama/27863)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ff38b98e5af6f28329e442ebbd4810a64c780867/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ff38b98e5af6f28329e442ebbd4810a64c780867"><tt>ff38b98</tt></a> metal : add fa-vec tunings for M4 Pro (llama/27824)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b6571e4a55279381fd342c4c9f9ab715f57bd217/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b6571e4a55279381fd342c4c9f9ab715f57bd217"><tt>b6571e4</tt></a> ggml-hexagon: add HTP unary ops for ABS and LOG (llama/27786)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a0614d9ea446915afde48e82e0d6697773c92a18/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a0614d9ea446915afde48e82e0d6697773c92a18"><tt>a0614d9</tt></a> hex-unary: fix RMS_NORM_MUL weight-offset bugs for grouped/broadcast norms (llama/27798)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/852997152e4ea554922aac6540c12e112deeb4e4/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/852997152e4ea554922aac6540c12e112deeb4e4"><tt>8529971</tt></a> opencl: add bin kernels <code>kernel_gemm_moe_q4_0_q8_1_dp4a_bin</code>, <code>kernel_gemm_moe_mxfp4_q8_1_dp4a_bin</code> (llama/27768)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/55ab1e517cf52c9441119f740bc0e9d758a4f4ec/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/55ab1e517cf52c9441119f740bc0e9d758a4f4ec"><tt>55ab1e5</tt></a> Feature: Added LIGHTNING_INDEXER support for Deepseek V4 ops on Vulkan Backend (llama/27453)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a5db1d66f0d32e7b5031d51f00797573f3ac7e02/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a5db1d66f0d32e7b5031d51f00797573f3ac7e02"><tt>a5db1d6</tt></a> metal : fix memory leaks due to missing autoreleasepools (llama/27758)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3fea10db5435faa9f19d95adedbe6963b2d0b45d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3fea10db5435faa9f19d95adedbe6963b2d0b45d"><tt>3fea10d</tt></a> hexagon: support for multi-NPU devices (IQ9, IQ10) and fully asynchronous backend (llama/26501)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5271734e0e4e1761c601dc4fcf41a877e22d3a80/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5271734e0e4e1761c601dc4fcf41a877e22d3a80"><tt>5271734</tt></a> vulkan: warptiles currently assume warp sizes <= 64, clamp to work around larger warps (llama/27726)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0a026972dae2f58ac5742e750767ac8e159b16e1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0a026972dae2f58ac5742e750767ac8e159b16e1"><tt>0a02697</tt></a> Implemented vulkan cross_entropy_loss and cross_entropy_loss_back (llama/27216)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/82f5f85e951855cdfa3f91578be6b1fe9a56c0bc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/82f5f85e951855cdfa3f91578be6b1fe9a56c0bc"><tt>82f5f85</tt></a> rpc : implement event and async backend APIs (llama/18626)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/9d8e6b91b18a15728c3c57b8eef30e16a413112f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/9d8e6b91b18a15728c3c57b8eef30e16a413112f"><tt>9d8e6b9</tt></a> cuda: unblock mmq for MoE on sm_60 (llama/26264)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8c0adb05feb14f2ac29a9b3d6856d3aff39a1596/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8c0adb05feb14f2ac29a9b3d6856d3aff39a1596"><tt>8c0adb0</tt></a> ggml-metal: add chunked SSD MMA for Mamba-2 prefill optimization (llama/26647)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8df657a2fa88a9778c99300ac7e2a52734ed57e3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8df657a2fa88a9778c99300ac7e2a52734ed57e3"><tt>8df657a</tt></a> ggml-meta: propagate buffer usage and call init on the new tensors (llama/27586)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/482956e744ea8484256985592d2018b2940dc4bc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/482956e744ea8484256985592d2018b2940dc4bc"><tt>482956e</tt></a> kleidiai: Rework KleidiAI Build System/Integration (llama/26077)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/e820c280a64b37c14e9c3aa4879dda51451ecb72/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/e820c280a64b37c14e9c3aa4879dda51451ecb72"><tt>e820c28</tt></a> rpc: support apple RDMA as an RPC transport (llama/26421)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/be12d39a17de113be24f3433c802d9c9ac044eb3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/be12d39a17de113be24f3433c802d9c9ac044eb3"><tt>be12d39</tt></a> metal : null-check buffer alloc to fix OOM crash (llama/25371)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/642b5d3260e020c2fc6f34a9569d10ddd7672963/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/642b5d3260e020c2fc6f34a9569d10ddd7672963"><tt>642b5d3</tt></a> ruby : Add #free method, check MemoryView strictly (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5337915002" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4032" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4032/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4032">#4032</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/eacbd8234c6654cdbf2c377f72b2106875479bdc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/eacbd8234c6654cdbf2c377f72b2106875479bdc"><tt>eacbd82</tt></a> whisper : default-initialize whisper_mel to avoid uninitialized read (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5104861699" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3981" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3981/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3981">#3981</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c4ac0012a8f5a2082dfca6aad4ddfd8b2c02b337/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c4ac0012a8f5a2082dfca6aad4ddfd8b2c02b337"><tt>c4ac001</tt></a> parakeet : fix TDT decode by outputting raw logits from the joint graph (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5246196097" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4017" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4017/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4017">#4017</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/978113305b2ead22249b881deafa131dc8884911/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/978113305b2ead22249b881deafa131dc8884911"><tt>9781133</tt></a> talk-llama : sync llama.cpp<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3680f66fa41c8ae781d75c5847ae87e525b48dd3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3680f66fa41c8ae781d75c5847ae87e525b48dd3"><tt>3680f66</tt></a> pi : init<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0414519d7f51338fd5669c49f1c78c31480b5245/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0414519d7f51338fd5669c49f1c78c31480b5245"><tt>0414519</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d470c9d4695b8decd29d90020c41733d1a7d8d87/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d470c9d4695b8decd29d90020c41733d1a7d8d87"><tt>d470c9d</tt></a> ggml : bump version to 0.22.0 (ggml/1607)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/322a77cf6d2a70cc5af500e83a2acf9e8ab40e01/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/322a77cf6d2a70cc5af500e83a2acf9e8ab40e01"><tt>322a77c</tt></a> sycl : mark tq2_0 as not supported (llama/27660)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/17a522a8cc1494436641e716fd2e8ea9dd7180e4/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/17a522a8cc1494436641e716fd2e8ea9dd7180e4"><tt>17a522a</tt></a> webgpu : fix handling of infinity values during ARGSORT and TOP_K (llama/27538)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fa3b87c1882c914ca3bd93a068c2e61a5196ecd0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fa3b87c1882c914ca3bd93a068c2e61a5196ecd0"><tt>fa3b87c</tt></a> metal : per-device tuned (Q, NE) for flash-attn vec (llama/26570)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fe52277c1fe56665a0bcb03c2eb2ec3b2e0d25c6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fe52277c1fe56665a0bcb03c2eb2ec3b2e0d25c6"><tt>fe52277</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b15d31d0fdcb11dc02be17fbfc7e54a69d4b0c1c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b15d31d0fdcb11dc02be17fbfc7e54a69d4b0c1c"><tt>b15d31d</tt></a> metal: per-op source split + parallel compile (llama/26561)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4257445149dafbff68e5125a79844f038590016b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4257445149dafbff68e5125a79844f038590016b"><tt>4257445</tt></a> scripts : update ggml-am<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1d8e05291a009236881b0d19c5dc631b47cd14eb/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1d8e05291a009236881b0d19c5dc631b47cd14eb"><tt>1d8e052</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/aa25d33f58e52be8dd0fedfc86224ea34a5371dc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/aa25d33f58e52be8dd0fedfc86224ea34a5371dc"><tt>aa25d33</tt></a> ggml : shorten virtual device naming in CUDA and Metal (llama/27608)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/103305e6560f0c9ac8bfd03343529af2aef6930f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/103305e6560f0c9ac8bfd03343529af2aef6930f"><tt>103305e</tt></a> webgpu : reorder includes since V that appears in common_decls.tmpl may be defined as K in flash_attn_decls.tmpl if KV_OVERLAP (llama/27545)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/32d0f49ddc84f72ca92f1fe04e660c3a7e7f1df3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/32d0f49ddc84f72ca92f1fe04e660c3a7e7f1df3"><tt>32d0f49</tt></a> ggml : fix ggml_clamp (llama/27644)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/20209c25f726ad964106812931a1fb9453df7ca9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/20209c25f726ad964106812931a1fb9453df7ca9"><tt>20209c2</tt></a> Deepseek 4: <code>-sm tensor</code> (llama/26490)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3b89b37c3094b5ed99e6500eef7ff7ee71a301de/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3b89b37c3094b5ed99e6500eef7ff7ee71a301de"><tt>3b89b37</tt></a> Fix meta tensor split state propagation (llama/27574)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c8a40099fa214d2210c18f404c40034bb127a91d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c8a40099fa214d2210c18f404c40034bb127a91d"><tt>c8a4009</tt></a> cuda : add POOL_1D support (llama/27573)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/13a7856daa5437becfa9000cd09335bbfcf1c5b1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/13a7856daa5437becfa9000cd09335bbfcf1c5b1"><tt>13a7856</tt></a> vulkan : added the PAD_REFLECT_1D operation (llama/26586)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1efb31e6fdd01f7fa4d9742c686f1724a973f252/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1efb31e6fdd01f7fa4d9742c686f1724a973f252"><tt>1efb31e</tt></a> ggml: optimize concat op by replacing per-element memcpy with row-level memcpy (llama/24575)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/21a67dd8dade100684b61d1ecc9fd1b4c67828b6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/21a67dd8dade100684b61d1ecc9fd1b4c67828b6"><tt>21a67dd</tt></a> sycl : add Q2_K reordered MMVQ and ESIMD kernels (again) (llama/27490)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5f7bd9ddb9ae3349e7fb57ef568a40b8cff1b975/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5f7bd9ddb9ae3349e7fb57ef568a40b8cff1b975"><tt>5f7bd9d</tt></a> opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (llama/26431)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a722846cb60859d0ed140559df3b45a2c2212a88/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a722846cb60859d0ed140559df3b45a2c2212a88"><tt>a722846</tt></a> whisper : guard null source in buffer loader read callback (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5104905936" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3982" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3982/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3982">#3982</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c122757fddf358397bb7f33b6ac3aab24a5bca04/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c122757fddf358397bb7f33b6ac3aab24a5bca04"><tt>c122757</tt></a> docs : center badges in README.md [no ci] (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5232057264" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4012" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4012/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4012">#4012</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/25694098b85937a98b826db79cbce38a87a97cf4/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/25694098b85937a98b826db79cbce38a87a97cf4"><tt>2569409</tt></a> devops : add main-rocm Dockerfile (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5081560806" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3975" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3975/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3975">#3975</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/52dec9d889cc8e5f0a1978f2eb37a96978a131be/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/52dec9d889cc8e5f0a1978f2eb37a96978a131be"><tt>52dec9d</tt></a> vitisai : add VitisAI Plugin for AMD Ryzen AI NPU encoder offload (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="3810806730" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3608" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3608/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3608">#3608</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/233fe1fc9b48a09e361d3594520838ca266537fe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/233fe1fc9b48a09e361d3594520838ca266537fe"><tt>233fe1f</tt></a> whisper : bypass cross-attention scaling for OpenVINO backend (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5192225621" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3997" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3997/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3997">#3997</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/51de5e8bb0c4e140092e84742e9fa0c2f56d8d62/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/51de5e8bb0c4e140092e84742e9fa0c2f56d8d62"><tt>51de5e8</tt></a> openvino : update model conversion and README.md (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5211548110" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4003" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4003/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4003">#4003</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3391d6b6f997f9e1d8f5f2c4e188c5dfe12b4942/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3391d6b6f997f9e1d8f5f2c4e188c5dfe12b4942"><tt>3391d6b</tt></a> scripts : add release.sh script (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5214614737" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4010" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4010/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4010">#4010</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a4610c78bf2fc0c6c0538612a8e44308ee2c25e5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a4610c78bf2fc0c6c0538612a8e44308ee2c25e5"><tt>a4610c7</tt></a> docs : add release badge and remove stable/roadmap [no ci] (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5214438089" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4009" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4009/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4009">#4009</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ab578879ecba93490fb17c7a57ca2d376088b5ae/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ab578879ecba93490fb17c7a57ca2d376088b5ae"><tt>ab57887</tt></a> make : add --parallel to cmake build command (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5214041106" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4007" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4007/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4007">#4007</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/45f1593fd326b3435c04392e3151dff65967e523/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/45f1593fd326b3435c04392e3151dff65967e523"><tt>45f1593</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ce77728c06fa6b5898f8ef557768c2fe3e129d0a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ce77728c06fa6b5898f8ef557768c2fe3e129d0a"><tt>ce77728</tt></a> Revert "sycl : add Q2_K reordered MMVQ and ESIMD kernels (llama/26336)" (llama/27486)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0d9ba28cdea41090569ae6734dd27dce047c8412/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0d9ba28cdea41090569ae6734dd27dce047c8412"><tt>0d9ba28</tt></a> ggml : bump version to 0.21.0 (ggml/1597)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d6c416e22266ac6d2ff404e589bbe5f1ea5a4f41/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d6c416e22266ac6d2ff404e589bbe5f1ea5a4f41"><tt>d6c416e</tt></a> kleidiai : add SME2 F32 GEMV kernel support (llama/26891)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d60ef65077d01ebbc5a83733231d496cccdc91f8/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d60ef65077d01ebbc5a83733231d496cccdc91f8"><tt>d60ef65</tt></a> sycl : add Q2_K reordered MMVQ and ESIMD kernels (llama/26336)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b1cb80596581107941184ec7e1b8bbdee7ad30d1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b1cb80596581107941184ec7e1b8bbdee7ad30d1"><tt>b1cb805</tt></a> sycl : Add Q5_K ESIMD kernel (llama/26376)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2cb52ddbc0e7e5615e4362e754a8511965f402f2/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2cb52ddbc0e7e5615e4362e754a8511965f402f2"><tt>2cb52dd</tt></a> opencl: keep the vocab-scale K-quant lm_head on the CPU for Adreno A7X (compiler issue workaround) (llama/26440)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/af74f979654d57a6d15469c220b94926a09d3858/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/af74f979654d57a6d15469c220b94926a09d3858"><tt>af74f97</tt></a> sycl: Update gate logic for Alchemist GPUs regarding OneDNN features. (llama/26635)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f19250e5acff1242e5588490ba7351380537a814/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f19250e5acff1242e5588490ba7351380537a814"><tt>f19250e</tt></a> sycl: fix multiple warnings in compiling sycl backend (llama/26713)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/12137c316d7d90ab2cb9e8fc1a580cd1cff9ad31/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/12137c316d7d90ab2cb9e8fc1a580cd1cff9ad31"><tt>12137c3</tt></a> sycl : fix load model with mlock issue (llama/27250)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5656e44ea26c3d9b38dc708410d35ebc6ac8cad5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5656e44ea26c3d9b38dc708410d35ebc6ac8cad5"><tt>5656e44</tt></a> ggml: support ggml_rope_set_offset on opencl, sycl, wgpu, hexagon (llama/27345)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d68216aa3ece8bf56a3ddc80a0a238e09028ab8c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d68216aa3ece8bf56a3ddc80a0a238e09028ab8c"><tt>d68216a</tt></a> metal : clamp K extent in tensor API mat-mat kernel for K not a multiple of 32 (llama/27450)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c16cb42f540885ac9cc44b0387d0ef81dea7f13e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c16cb42f540885ac9cc44b0387d0ef81dea7f13e"><tt>c16cb42</tt></a> opencl: fix q6_K flat mul_mat for Adreno A6x/A7x GPUs with older E031 compilers (llama/26476)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/73c2b7ec6f4abb3b6c3130a21fbfb3f72c3735bd/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/73c2b7ec6f4abb3b6c3130a21fbfb3f72c3735bd"><tt>73c2b7e</tt></a> opencl: fix local size for norm (llama/27339)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8539d72ffab645c5e0eb3c3436850ab117c2d6f2/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8539d72ffab645c5e0eb3c3436850ab117c2d6f2"><tt>8539d72</tt></a> vulkan: FA MMQ should use fp32 for Q quantization calculations (llama/27413)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/60f14a5eccc38a4c71e59b254b366400a84ab167/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/60f14a5eccc38a4c71e59b254b366400a84ab167"><tt>60f14a5</tt></a> metal : dequant kv cache only for large batches (llama/27438)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/292241818ac77a30e58b414213007cd3f24ed430/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/292241818ac77a30e58b414213007cd3f24ed430"><tt>2922418</tt></a> CI: Use LLVM's OpenMP over MSVC_DEBUG_non_redist on Windows (llama/26678)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c2dc946f3f548c78f2a32c6e0efbbff477b0ff8a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c2dc946f3f548c78f2a32c6e0efbbff477b0ff8a"><tt>c2dc946</tt></a> CUDA: adding switch points per HW and quant type to tune the mvq->MMQ decode crossover (llama/26079)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/283775e0a5deee4a212287b2858039385ab1dd2e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/283775e0a5deee4a212287b2858039385ab1dd2e"><tt>283775e</tt></a> metal : dequantize quantized KV to F16 before flash attention (llama/27390)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/02be8f5da8955883a86787b83eb8272e6ecdc80a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/02be8f5da8955883a86787b83eb8272e6ecdc80a"><tt>02be8f5</tt></a> Revert "tensor-split meta backend fixes (ggml/26502)" (llama/27433)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/13fa860764e6c8f1fa20165c99fe7295e78dff24/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/13fa860764e6c8f1fa20165c99fe7295e78dff24"><tt>13fa860</tt></a> ggml: fix backend split scheduler race condition (llama/26040)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2648a706e0eeadd7c5720aa31df2fbddc2f473d6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2648a706e0eeadd7c5720aa31df2fbddc2f473d6"><tt>2648a70</tt></a> ggml-cuda: provide static workspace for cuBLAS handles (llama/26574)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/acfad328414fb48d51b2d2394fe67d3df187af3f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/acfad328414fb48d51b2d2394fe67d3df187af3f"><tt>acfad32</tt></a> vulkan : add source groups for shaders (llama/26666)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a3386d40e907a56f1ad1642a41f70add897f3e70/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a3386d40e907a56f1ad1642a41f70add897f3e70"><tt>a3386d4</tt></a> opencl: make the MoE expert scatter deterministic (llama/26464)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/135f348bc46c80254f379200945eeede74dbd182/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/135f348bc46c80254f379200945eeede74dbd182"><tt>135f348</tt></a> tensor-split meta backend fixes (llama/26502)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/9f4b18ae6e45602ad13417b627bb3d5bf9f330eb/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/9f4b18ae6e45602ad13417b627bb3d5bf9f330eb"><tt>9f4b18a</tt></a> hexagon: fix FA HMX queue ordering and pack the rescale D matrices (llama/27042)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/cd340ea8320751fdd213d5dca56f4bde1ebfbac2/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/cd340ea8320751fdd213d5dca56f4bde1ebfbac2"><tt>cd340ea</tt></a> opencl: port fused ssm_scan kernel (Mamba-2, d_state in {128, 256}) to GPU (llama/26439)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3d277427a72954d82380191609442d1452aace47/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3d277427a72954d82380191609442d1452aace47"><tt>3d27742</tt></a> ggml-cpu: gate __fp16 on __ARM_FP16_FORMAT_IEEE (llama/26860)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/11e58f3edba075fe30cef5aab44ca59c7fa606df/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/11e58f3edba075fe30cef5aab44ca59c7fa606df"><tt>11e58f3</tt></a> vulkan : dequant q8_0 KV once in coopmat1 (llama/25494)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4ef3e871cad0ba96c964c571c96a1b87694c7bcd/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4ef3e871cad0ba96c964c571c96a1b87694c7bcd"><tt>4ef3e87</tt></a> vulkan: add null checks in ggml_vk_queue_command_pools_cleanup (llama/27353)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/689ad69f035c1c21e636c28ef93f5c8315634d85/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/689ad69f035c1c21e636c28ef93f5c8315634d85"><tt>689ad69</tt></a> sycl: report zero devices instead of aborting when the host has none (llama/27291)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8442c74f5673972eb6feea583cd16d93a6d02a65/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8442c74f5673972eb6feea583cd16d93a6d02a65"><tt>8442c74</tt></a> ggml: add ggml_rope_set_offset (+ metal support) (llama/27120)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d830bd220cd0d8b217a3fc9b023ba5692ac4fba1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d830bd220cd0d8b217a3fc9b023ba5692ac4fba1"><tt>d830bd2</tt></a> metal : dequantize q8_0 using packed types (llama/27370)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1c882a8ed089b452f35b42ee099089662a6262f1/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1c882a8ed089b452f35b42ee099089662a6262f1"><tt>1c882a8</tt></a> vulkan: tiled transpose for 0<->2 permuted CONT (llama/26585)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/7df5fa8e62fa96bbfd7caf2eecb9252f84930c2c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/7df5fa8e62fa96bbfd7caf2eecb9252f84930c2c"><tt>7df5fa8</tt></a> ggml-webgpu: add mulmat with overlapping src0/src1 (e.g., for minimax-01) (llama/27321)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fa1e2bcaafcd33182f3f8d46b11683466ccf0781/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fa1e2bcaafcd33182f3f8d46b11683466ccf0781"><tt>fa1e2bc</tt></a> opencl: fix WAR race in the generic FA tile kernels when the WG spans subgroups (llama/26434)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4be3101670be56b2f5c4df303163172ea3f3bb0d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4be3101670be56b2f5c4df303163172ea3f3bb0d"><tt>4be3101</tt></a> RPC: populate use_count to enable fusion inside backends (llama/27142)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/dbab353f03595e658cd3e004e9258f97b5603c99/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/dbab353f03595e658cd3e004e9258f97b5603c99"><tt>dbab353</tt></a> sycl: honor GGML_HINT_SRC0_IS_HADAMARD (llama/27298)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c6a2bd0973ddd61bbc185f053d78c718f9c7beb9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c6a2bd0973ddd61bbc185f053d78c718f9c7beb9"><tt>c6a2bd0</tt></a> devops : stop shadowing proper cuda libraries in runtime docker build (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5144850603" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3989" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3989/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3989">#3989</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ec737355d280226b04042cbce055de842df87446/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ec737355d280226b04042cbce055de842df87446"><tt>ec73735</tt></a> tests: add check for nullptr for wctx in test-vad-full (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5196160712" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3998" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3998/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3998">#3998</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a4ad15f302cdb3478c40712644fd1883a098d3e4/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a4ad15f302cdb3478c40712644fd1883a098d3e4"><tt>a4ad15f</tt></a> ci : release clean-up (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5212205488" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4005" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4005/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4005">#4005</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/81a3fadf614179fb39975db7564cbfecd42fb92d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/81a3fadf614179fb39975db7564cbfecd42fb92d"><tt>81a3fad</tt></a> server : only enable token timestamps when the response needs them (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5146484197" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3990" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3990/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3990">#3990</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d61695dd87c21d5e90186e808f3b95778241c4ba/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d61695dd87c21d5e90186e808f3b95778241c4ba"><tt>d61695d</tt></a> docs : fix typos in readme files (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5211817638" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4004" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4004/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4004">#4004</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b78df3d1dfbbe3093dd7040969d1f6713a715cd7/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b78df3d1dfbbe3093dd7040969d1f6713a715cd7"><tt>b78df3d</tt></a> ci : move close-issue.yml to .github/workflows [no ci] (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5211119884" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4002" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4002/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4002">#4002</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/339f2b4e27d7c3b52f44a124a854abba507acff3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/339f2b4e27d7c3b52f44a124a854abba507acff3"><tt>339f2b4</tt></a> bindings-javascript : remove package.json from git (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5203867393" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4001" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4001/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4001">#4001</a>)</p> github-actions[bot] tag:github.com,2008:Repository/541269386/b5130 2026-09-11T05:30:01Z

b5130

<p>metal : remove leftover ggml-metal.metal kernels file (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/whisper.cpp/pull/4051">#4051</a>)</p> <p>This commit removes the ggml-metal.metal file that used to contain all <br />the metal kernels. These have now been separated into separate kernels <br />in ggml/src/ggml-metal/kernels and it seems that this file was not <br />removed when synced with ggml.</p> <p>This is currently preventing whisper.cpp to be released as the release <br />check fails.</p> github-actions[bot] tag:github.com,2008:Repository/541269386/b5127 2026-09-10T11:56:28Z

b5127

<p><strong>Note:</strong> the Windows arm64 (CUDA) build (<code>whisper-bin-win-cuda-*-arm64.zip</code>) uses a preview edition of the CUDA Toolkit for Windows on Arm and should be considered experimental.</p> github-actions[bot] tag:github.com,2008:Repository/541269386/v1.9.3 2026-08-20T11:42:23Z

v1.9.3

<div class="markdown-alert markdown-alert-note"><p class="markdown-alert-title"><svg data-component="Octicon" class="octicon octicon-info mr-2" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="M0 8a8 8 0 1 1 16 0A8 8 0 0 1 0 8Zm8-6.5a6.5 6.5 0 1 0 0 13 6.5 6.5 0 0 0 0-13ZM6.5 7.75A.75.75 0 0 1 7.25 7h1a.75.75 0 0 1 .75.75v2.75h.25a.75.75 0 0 1 0 1.5h-2a.75.75 0 0 1 0-1.5h.25v-2h-.25a.75.75 0 0 1-.75-.75ZM8 6a1 1 0 1 1 0-2 1 1 0 0 1 0 2Z"></path></svg>Note</p><p>Semantic versioning is still work in progress.<br> More info can be found in <a href="https://github.com/ggml-org/ggml/discussions/1579">https://github.com/ggml-org/ggml/discussions/1579</a></p> </div> <p><strong>Nightly build:</strong> <a href="https://github.com/ggml-org/whisper.cpp/releases/tag/b4938">b4938</a></p> <h2>Change log since v1.9.2</h2> <p><a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/371b5a7561823ab2bb32142d2751e35e7534727b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/371b5a7561823ab2bb32142d2751e35e7534727b"><tt>371b5a7</tt></a> release : v1.9.3 (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5201276647" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/4000" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/4000/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/4000">#4000</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/81c190560a718ddf69b3ab2e7e9808b2bfb7842c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/81c190560a718ddf69b3ab2e7e9808b2bfb7842c"><tt>81c1905</tt></a> cmake : update semver and release process [no ci] (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5183411332" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3996" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3996/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3996">#3996</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4834a2327d008ace3ec5a9ed00f51454bcabbc1c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4834a2327d008ace3ec5a9ed00f51454bcabbc1c"><tt>4834a23</tt></a> talk-llama : sync llama.cpp<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/6b014cfbf6f69c70a0cf779bffb5fc58311f5fab/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/6b014cfbf6f69c70a0cf779bffb5fc58311f5fab"><tt>6b014cf</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8189458df8aac9b08966007935d65b806c322015/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8189458df8aac9b08966007935d65b806c322015"><tt>8189458</tt></a> ggml : bump version to 0.20.2 (ggml/1589)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/51319a2cd902ed0a67c96ac29db7219c48d17e35/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/51319a2cd902ed0a67c96ac29db7219c48d17e35"><tt>51319a2</tt></a> CUDA: MMVQ nwarps=8 for bs=1 for dense models on DGX Spark (llama/26843)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/71759b7c9f0418c10ecdf4927bc0f11490207bbe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/71759b7c9f0418c10ecdf4927bc0f11490207bbe"><tt>71759b7</tt></a> cuda : skip UMA override for HIP builds (llama/27083)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/964bb1b8f591e4917dba049b3228eec53836ddf3/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/964bb1b8f591e4917dba049b3228eec53836ddf3"><tt>964bb1b</tt></a> ggml : bump version to 0.20.1 (ggml/1587)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/4257f47019d27b7a5ff10b0d6f3ae4e43482415d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/4257f47019d27b7a5ff10b0d6f3ae4e43482415d"><tt>4257f47</tt></a> sycl: fix thread/block count in quantized cpy kernel launches (llama/27160)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0f3861385f9f9d980c4c13d7e1b2dd7e9502560d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0f3861385f9f9d980c4c13d7e1b2dd7e9502560d"><tt>0f38613</tt></a> support OP OPT_STEP_ADAMW, OPT_STEP_SGD (llama/25268)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/9a0d190241a97f18632432593e0f4bfd2bff6b76/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/9a0d190241a97f18632432593e0f4bfd2bff6b76"><tt>9a0d190</tt></a> vulkan: add SHMEM_STRIDE_PAD/APPLY_SLM_A_RESHAPE for coopmat1 on Intel Xe (llama/25380)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ab71410c08226a6e830ed6e935518fbbbf57b57e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ab71410c08226a6e830ed6e935518fbbbf57b57e"><tt>ab71410</tt></a> fixed indent<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/9b98e57d2b00a20719c0bda0e3910d11f72277eb/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/9b98e57d2b00a20719c0bda0e3910d11f72277eb"><tt>9b98e57</tt></a> Fixed gating logic for problematic Intel driver version<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1fe009caeda75f69bc864d6370b10674e45a92bd/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1fe009caeda75f69bc864d6370b10674e45a92bd"><tt>1fe009c</tt></a> talk-llama : fix build (#0)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/733f2819877b78987942653c5cd98a70169ed17e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/733f2819877b78987942653c5cd98a70169ed17e"><tt>733f281</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/667da04aa41946912feacf62bfad245b3b87f4aa/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/667da04aa41946912feacf62bfad245b3b87f4aa"><tt>667da04</tt></a> ggml : bump version to 0.20.0 (ggml/1584)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b7ea8b19bf387e1023ce8f86b9e090cd1faed92c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b7ea8b19bf387e1023ce8f86b9e090cd1faed92c"><tt>b7ea8b1</tt></a> ggml : recurrent state rollback for ggml_ssm_scan (llama/26623)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2aef2a0085f115d45c2a4adf5e706c42f070c854/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2aef2a0085f115d45c2a4adf5e706c42f070c854"><tt>2aef2a0</tt></a> sycl: fuse mul_mat(gate) + mul_mat(up) + GLU for q4_K dense FFN (llama/26779)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5bc0c8525445045d35e013dcc26d57d20f6283e6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5bc0c8525445045d35e013dcc26d57d20f6283e6"><tt>5bc0c85</tt></a> ggml: force single thread on wasi (llama/25686)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/2d05b6ec1ef8d41c47bcb8491c5828503f8c5cbd/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/2d05b6ec1ef8d41c47bcb8491c5828503f8c5cbd"><tt>2d05b6e</tt></a> sycl: fuse the gated-delta-net state writeback cpy (llama/26643)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/43cbe4157488574cf1d40997a00f079811177dea/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/43cbe4157488574cf1d40997a00f079811177dea"><tt>43cbe41</tt></a> OpenVINO: Qwen3.5, memory optimization, and test-recurrent-state-rollback (llama/26952)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/94d2d35a7470e2f6df4fa42c6d19b8e9a142091b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/94d2d35a7470e2f6df4fa42c6d19b8e9a142091b"><tt>94d2d35</tt></a> Support host pinned mem to improve SYCL Host-to-Device Memory Access (llama/26789)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/34009e8518a42b3bedf413ff9de4fd7078bd3f69/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/34009e8518a42b3bedf413ff9de4fd7078bd3f69"><tt>34009e8</tt></a> metal : add TQ2_0 support (llama/26980)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/62031fed0b4a2f16ba4fc3bb4567cfca19d89dfa/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/62031fed0b4a2f16ba4fc3bb4567cfca19d89dfa"><tt>62031fe</tt></a> ggml-cpu/ops: vectorize flash-attention V-cache F16 to F32 conversion (llama/26947)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/ac9a74f1a993bfd086666689aec7c3d16920ce6b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/ac9a74f1a993bfd086666689aec7c3d16920ce6b"><tt>ac9a74f</tt></a> sycl: remove separate fp32 type promotion in gemm non-oneDNN path (llama/26372)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/3425d130ee8651ffe88182905172f458f0760134/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/3425d130ee8651ffe88182905172f458f0760134"><tt>3425d13</tt></a> sycl: fuse UNARY(silu|sigmoid|softplus) + MUL (llama/26411)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/90d4ed185961b76ea411148c34a7653fc431752a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/90d4ed185961b76ea411148c34a7653fc431752a"><tt>90d4ed1</tt></a> sycl : Add DMMV ESIMD Q3_K kernel (llama/26251)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f7011606ad0d84205104556eb2b34de0d6a73364/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f7011606ad0d84205104556eb2b34de0d6a73364"><tt>f701160</tt></a> sycl : enhance concat to support Q4_0, Q4_1, Q5_0, Q5_1, Q8_0 (llama/26800)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/c7be8b3358c9cfcc00c35e94a899decae072f9ff/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/c7be8b3358c9cfcc00c35e94a899decae072f9ff"><tt>c7be8b3</tt></a> ggml-hip : remove -funsafe-math-optimizations (llama/26696)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/406b1168137083cb4907ce1060eb080a3409ff58/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/406b1168137083cb4907ce1060eb080a3409ff58"><tt>406b116</tt></a> ggml : fix arm builds, unused var (llama/26991)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b0e3297169ca5a3e968776057de57727be955a16/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b0e3297169ca5a3e968776057de57727be955a16"><tt>b0e3297</tt></a> gguf : harden loader against malformed tensor dims and metadata types (llama/25596)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/92423afd20f26eda9b8421ceee74e34da80119a0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/92423afd20f26eda9b8421ceee74e34da80119a0"><tt>92423af</tt></a> kleidiai: Add runtime feature detection mechanism for aarch64/kleidiai (llama/26076)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/1b93067bf9116a9b451ad206720fff6bc66e67f6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/1b93067bf9116a9b451ad206720fff6bc66e67f6"><tt>1b93067</tt></a> opencl: default FA c8 cluster width to 16 on X1E (llama/26433)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/632904002ddb0fa9f9731d287ada96fa5f54f11f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/632904002ddb0fa9f9731d287ada96fa5f54f11f"><tt>6329040</tt></a> vulkan: add TQ2_0 (ternary) support (llama/25850)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/db3687b2979a116a4556d8478e284f28080d8e6c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/db3687b2979a116a4556d8478e284f28080d8e6c"><tt>db3687b</tt></a> opencl: use flat mv q5_k when weight exceeds image1d_buffer_t limit (llama/26880)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/cdbe4551023d96ebc906d814798dc4a3d0fe75e9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/cdbe4551023d96ebc906d814798dc4a3d0fe75e9"><tt>cdbe455</tt></a> CUDA: only disable CUDA graphs when mul_mat_id actually needs a stream sync (llama/26802)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/030656f26f139e7d94c92942d40424417046ca36/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/030656f26f139e7d94c92942d40424417046ca36"><tt>030656f</tt></a> cuda : add warp-per-row wkv7 kernel for single-token decode (llama/26111)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/d9589688cbd4dd3a9e81c879cf72ff5a4eafcf40/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/d9589688cbd4dd3a9e81c879cf72ff5a4eafcf40"><tt>d958968</tt></a> llama: add default load-mode auto, which avoids mmap on iGPUs (llama/26081)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/0f60f8e691c5db8fdd673329d3258254e2c76604/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/0f60f8e691c5db8fdd673329d3258254e2c76604"><tt>0f60f8e</tt></a> ggml-webgpu: fix CI errors from #25025 and #25262 (llama/26566)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/a6e2630ff142cd97289bdee6224855a950de38a0/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/a6e2630ff142cd97289bdee6224855a950de38a0"><tt>a6e2630</tt></a> opencl: transpose the K tile in local memory for FA prefill kernels (llama/26428)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/830ec22572404776ed3857d9b2424a90508b0806/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/830ec22572404776ed3857d9b2424a90508b0806"><tt>830ec22</tt></a> ggml-cpu : fix CPU affinity mask being ignored on Android (llama/26838)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/b3bc90463807a97763c2f1e4082bd9925f4dbc7f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/b3bc90463807a97763c2f1e4082bd9925f4dbc7f"><tt>b3bc904</tt></a> ggml : require contiguous src for ROLL on CUDA and Metal (llama/25928)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/f09a97cf6dddc821bc029c6043cd061aff8be66d/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/f09a97cf6dddc821bc029c6043cd061aff8be66d"><tt>f09a97c</tt></a> ggml-webgpu : refactor several wgsl files and simplify flash_attn wgsl. (llama/26134)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/877761c5b6d2429a5559b5566c4c285f53a50c04/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/877761c5b6d2429a5559b5566c4c285f53a50c04"><tt>877761c</tt></a> ggml-cpu : fix missing Q5_0 dispatch in SpaceMiT backend (llama/26792)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/10791af67498011a99d361f02d3684e5d3b5c0c9/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/10791af67498011a99d361f02d3684e5d3b5c0c9"><tt>10791af</tt></a> CUDA: fuse rms_norm + mul + rope (+ view + set_rows) (llama/26767)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/068d3b0b4530506a884fe88aa6e219e870697992/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/068d3b0b4530506a884fe88aa6e219e870697992"><tt>068d3b0</tt></a> CUDA: fix thread/block count in quantized cpy kernel launches (llama/26731)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/eb3296f278380600699b9921b136633e6d3a0abe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/eb3296f278380600699b9921b136633e6d3a0abe"><tt>eb3296f</tt></a> sycl: coalesce the ssm_conv window loads (llama/26612)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8a5ba01c93e6f3c6961a83bb345ed7bc5cc33f8a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8a5ba01c93e6f3c6961a83bb345ed7bc5cc33f8a"><tt>8a5ba01</tt></a> metal : fix NORM/RMS_NORM for row lengths that leave a partial simdgroup (llama/26708)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/06248ca1107bfe1e906dcf9e9657a9719f1f346c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/06248ca1107bfe1e906dcf9e9657a9719f1f346c"><tt>06248ca</tt></a> cmake : add config version support (ggml/1582)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/592feef04a1802b18cbeffd0fd0eb5d02570c2ec/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/592feef04a1802b18cbeffd0fd0eb5d02570c2ec"><tt>592feef</tt></a> talk-llama : sync llama.cpp<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8770492611beac58a79e8bbb90cbd32f09b42bdb/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8770492611beac58a79e8bbb90cbd32f09b42bdb"><tt>8770492</tt></a> sync : ggml<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/84cdcad3570cdc3cdcf08e920d9de45c6ddbbcdc/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/84cdcad3570cdc3cdcf08e920d9de45c6ddbbcdc"><tt>84cdcad</tt></a> ggml : bump version to 0.19.0 (ggml/1581)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8587ad3bd0537608e951ddec383f0069b37c2ea6/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8587ad3bd0537608e951ddec383f0069b37c2ea6"><tt>8587ad3</tt></a> ggml : add aarch64 HWCAP fallbacks and fix fp16 variant detection (llama/25554)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/56cb15472d5899256f1cfb9abe7b207bbdaa38fe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/56cb15472d5899256f1cfb9abe7b207bbdaa38fe"><tt>56cb154</tt></a> sycl: fix UE4M3 parsing (llama/25608)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/077c5d42d27f3f46658965d1f4afbb95e5ff1b1e/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/077c5d42d27f3f46658965d1f4afbb95e5ff1b1e"><tt>077c5d4</tt></a> sycl: *glu flat path (llama/26354)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/9faa9ee7b0e54828a21149e2e02bdd04123a696b/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/9faa9ee7b0e54828a21149e2e02bdd04123a696b"><tt>9faa9ee</tt></a> sycl : Support DSv4 OPs: LIGHTNING_INDEXER,DSV4_HC_COMB,DSV4_HC_POST,DSV4_HC_PRE (llama/26568)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/fb9e8ca97b8baad1eaa92926c433a184e0eadb6a/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/fb9e8ca97b8baad1eaa92926c433a184e0eadb6a"><tt>fb9e8ca</tt></a> sycl : fix error Error OP FLASH_ATTN_EXT on arc770 (llama/26441)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/89d45afe250477cf9567aecb0a1327ef97494ce7/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/89d45afe250477cf9567aecb0a1327ef97494ce7"><tt>89d45af</tt></a> sycl : enhance OP set_rows to support all missed data types (llama/26515)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/79ab70c9222a45443eda2df193293f4b1453578c/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/79ab70c9222a45443eda2df193293f4b1453578c"><tt>79ab70c</tt></a> cuda: fix warnings for unused variable/function (llama/26688)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5693378610931163156cfbb15dd221fdffb50954/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5693378610931163156cfbb15dd221fdffb50954"><tt>5693378</tt></a> metal : avoid <code>threadgroup</code> matrix array instantiation in kernel_lightning_indexer (llama/26646)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/69bd0a9ac5058585004d5063ac102dbd0427fe00/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/69bd0a9ac5058585004d5063ac102dbd0427fe00"><tt>69bd0a9</tt></a> ci : onboard AMD ROCm CI with gfx1151 fixes (llama/26544)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/5a80d0ad900ae2fd0dcb3bc2a38cafbb25e30246/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/5a80d0ad900ae2fd0dcb3bc2a38cafbb25e30246"><tt>5a80d0a</tt></a> vulkan: fix submission batching size, add debug tools for diagnosing causes of DeviceLost drivers errors (llama/26371)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/60894f1e3766ef0224b57ad8840a83a66e05cb6f/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/60894f1e3766ef0224b57ad8840a83a66e05cb6f"><tt>60894f1</tt></a> mtmd/ggml: add ggml_build_forward_order (llama/26649)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/873102193ab7238fcccb4a9169a522eebbe4bc00/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/873102193ab7238fcccb4a9169a522eebbe4bc00"><tt>8731021</tt></a> vulkan backend ops: implemented GATED_LINEAR_ATTN (llama/25601)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/8631825d41a2712268813981a9550b04a3f225e5/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/8631825d41a2712268813981a9550b04a3f225e5"><tt>8631825</tt></a> whisper : heap out-of-bounds read in log_mel_spectrogram on very short audio (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4981650094" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3956" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3956/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3956">#3956</a>)<br> <a class="commit-link" data-hovercard-type="commit" data-hovercard-url="https://github.com/ggml-org/whisper.cpp/commit/df1547b6fe362556e48fd21f386c6eee85b971fe/hovercard" href="https://github.com/ggml-org/whisper.cpp/commit/df1547b6fe362556e48fd21f386c6eee85b971fe"><tt>df1547b</tt></a> whisper,parakeet : reject invalid n_dims in tensor header to prevent stack-buffer-overflow on malformed model files (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4984472659" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3957" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3957/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3957">#3957</a>)</p> github-actions[bot] tag:github.com,2008:Repository/541269386/b4938 2026-08-20T11:33:53Z

b4938

No content. github-actions[bot] tag:github.com,2008:Repository/541269386/v1.9.2 2026-08-04T15:31:37Z

v1.9.2

<h2>What's Changed</h2> <ul> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4698744851" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3896" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3896/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3896">#3896</a></li> <li>include parakeet in build-xcframework.sh by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/daaku/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/daaku">@daaku</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4711374611" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3899" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3899/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3899">#3899</a></li> <li>Fix pkgconfig configuration (Nix build failure) by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/nmouha/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/nmouha">@nmouha</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4696104766" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3894" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3894/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3894">#3894</a></li> <li>examples : update model names in parakeet-cli README.md [no ci] by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4723450019" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3906" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3906/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3906">#3906</a></li> <li>examples: Fix argument flag for min speech duration in VAD by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/QuantiusBenignus/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/QuantiusBenignus">@QuantiusBenignus</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4728682676" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3907" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3907/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3907">#3907</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4751850024" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3912" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3912/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3912">#3912</a></li> <li>examples : Improved inference performance of Android example project by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/yymin1022/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/yymin1022">@yymin1022</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4753087386" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3913" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3913/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3913">#3913</a></li> <li>ci : set GGML_NATIVE=OFF for sanitize builds by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4777103565" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3920" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3920/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3920">#3920</a></li> <li>whisper : make voice_length() utf-8 aware for CJK by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/buxuku/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/buxuku">@buxuku</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4758497846" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3915" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3915/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3915">#3915</a></li> <li>whisper : map token timestamps to original time when VAD is enabled by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/buxuku/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/buxuku">@buxuku</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4742983977" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3910" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3910/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3910">#3910</a></li> <li>ci, examples : add package-lock.json to examples addon node by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4776615087" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3919" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3919/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3919">#3919</a></li> <li>whisper : expose internal VAD speech segments by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/buxuku/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/buxuku">@buxuku</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4758498629" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3916" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3916/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3916">#3916</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4853073864" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3940" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3940/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3940">#3940</a></li> <li>Fix <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="1616311451" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/587" data-hovercard-type="issue" data-hovercard-url="/ggml-org/whisper.cpp/issues/587/hovercard" href="https://github.com/ggml-org/whisper.cpp/issues/587">#587</a>: Remove leading space from txt output by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/JingliangGao/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/JingliangGao">@JingliangGao</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4782254491" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3921" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3921/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3921">#3921</a></li> <li>ruby : add VAD speech segments API by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/KitaitiMakoto/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/KitaitiMakoto">@KitaitiMakoto</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4819643279" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3931" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3931/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3931">#3931</a></li> <li>Adding ruby bindings change log by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/taf2/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/taf2">@taf2</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5004545227" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3959" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3959/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3959">#3959</a></li> <li>verify hparams loaded from parakeet model bin file by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/bhargav191098/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/bhargav191098">@bhargav191098</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4952886490" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3950" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3950/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3950">#3950</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5020284883" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3962" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3962/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3962">#3962</a></li> <li>Fix VAD min silence argument parsing by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/chupakobra6/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/chupakobra6">@chupakobra6</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5028922167" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3963" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3963/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3963">#3963</a></li> <li>ci : set GGML_NATIVE=OFF for build-vad workflow by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5046822953" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3966" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3966/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3966">#3966</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5058563060" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3969" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3969/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3969">#3969</a></li> <li>release : v1.9.2 by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5059650533" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3970" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3970/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3970">#3970</a></li> </ul> <h2>New Contributors</h2> <ul> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/daaku/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/daaku">@daaku</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4711374611" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3899" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3899/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3899">#3899</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/nmouha/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/nmouha">@nmouha</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4696104766" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3894" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3894/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3894">#3894</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/QuantiusBenignus/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/QuantiusBenignus">@QuantiusBenignus</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4728682676" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3907" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3907/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3907">#3907</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/yymin1022/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/yymin1022">@yymin1022</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4753087386" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3913" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3913/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3913">#3913</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/JingliangGao/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/JingliangGao">@JingliangGao</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4782254491" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3921" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3921/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3921">#3921</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/bhargav191098/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/bhargav191098">@bhargav191098</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4952886490" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3950" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3950/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3950">#3950</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/chupakobra6/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/chupakobra6">@chupakobra6</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5028922167" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3963" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3963/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3963">#3963</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ggml-org/whisper.cpp/compare/v1.9.1...v1.9.2"><tt>v1.9.1...v1.9.2</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/541269386/v1.9.1 2026-06-19T05:53:19Z

v1.9.1

<h2>What's Changed</h2> <ul> <li>ci : add GGML_NATIVE=OFF and GGML_BMI2=OFF to windows-blas by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4689145470" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3891" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3891/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3891">#3891</a></li> <li>release : v1.9.1 by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4692519381" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3892" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3892/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3892">#3892</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ggml-org/whisper.cpp/compare/v1.9.0...v1.9.1"><tt>v1.9.0...v1.9.1</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/541269386/v1.9.0 2026-06-17T10:19:21Z

v1.9.0

<h2>What's Changed</h2> <ul> <li>ci : only trigger release jobs for tags by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4673622529" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3883" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3883/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3883">#3883</a></li> <li>parakeet : add support for NVIDIA Parakeet by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4186000945" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3735" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3735/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3735">#3735</a></li> <li>ruby : add support for Parakeet by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/KitaitiMakoto/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/KitaitiMakoto">@KitaitiMakoto</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4679252865" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3885" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3885/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3885">#3885</a></li> <li>release : v1.9.0 by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4680762656" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3886" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3886/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3886">#3886</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ggml-org/whisper.cpp/compare/v1.8.7...v1.9.0"><tt>v1.8.7...v1.9.0</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/541269386/v1.8.7 2026-06-16T10:59:32Z

v1.8.7

<h2>What's Changed</h2> <ul> <li>Do not assume /usr/lib library installation. by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/waptaff/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/waptaff">@waptaff</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4032156986" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3693" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3693/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3693">#3693</a></li> <li>server : merge split UTF-8 token text in verbose JSON by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/lyonsno/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/lyonsno">@lyonsno</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4559414232" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3850" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3850/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3850">#3850</a></li> <li>whisper : catch C++ exceptions in whisper_init_with_params_no_state by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danscMax/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danscMax">@danscMax</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4534758305" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3831" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3831/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3831">#3831</a></li> <li>ci : refactor + optimize by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4557848939" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3847" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3847/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3847">#3847</a></li> <li>ci : only publish/push docker images daily by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4586506641" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3854" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3854/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3854">#3854</a></li> <li>ci : use ccache instead of sccache for windows-cublas [no ci] by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4587140802" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3855" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3855/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3855">#3855</a></li> <li>ci : pin github actions to commit SHAs by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4587356452" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3856" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3856/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3856">#3856</a></li> <li>ci : use emscripten-core and pin version by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4588321190" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3857" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3857/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3857">#3857</a></li> <li>ci: build-windows action slimming by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4589103991" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3858" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3858/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3858">#3858</a></li> <li>Ci build quantize vad wasm ccache by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4597111847" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3860" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3860/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3860">#3860</a></li> <li>ci : add HF_TOKEN to docker.yml workflow [no ci] by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4602242799" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3861" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3861/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3861">#3861</a></li> <li>ci : add ccache to build-sycl [no ci] by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4596327177" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3859" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3859/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3859">#3859</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4611837608" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3867" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3867/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3867">#3867</a></li> <li>coreml : fix --quantize crash and --optimize-ane label for mlprogram format by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/krystophny/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/krystophny">@krystophny</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4612278071" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3868" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3868/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3868">#3868</a></li> <li>ci : pin github actions to commit sha's by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4610319312" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3865" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3865/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3865">#3865</a></li> <li>sync : ggml by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4662560272" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3880" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3880/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3880">#3880</a></li> <li>cli : add <code>--version</code> flag by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/rumitvn/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/rumitvn">@rumitvn</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4659793708" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3878" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3878/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3878">#3878</a></li> <li>release : v1.8.7 by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4673051935" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3881" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3881/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3881">#3881</a></li> </ul> <h2>New Contributors</h2> <ul> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/waptaff/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/waptaff">@waptaff</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4032156986" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3693" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3693/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3693">#3693</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/lyonsno/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/lyonsno">@lyonsno</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4559414232" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3850" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3850/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3850">#3850</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danscMax/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danscMax">@danscMax</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4534758305" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3831" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3831/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3831">#3831</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/krystophny/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/krystophny">@krystophny</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4612278071" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3868" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3868/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3868">#3868</a></li> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/rumitvn/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/rumitvn">@rumitvn</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4659793708" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3878" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3878/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3878">#3878</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ggml-org/whisper.cpp/compare/v1.8.6...v1.8.7"><tt>v1.8.6...v1.8.7</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/541269386/v1.8.6 2026-06-02T06:22:27Z

v1.8.6

<h2>What's Changed</h2> <ul> <li>common : re-implement <code>ffmpeg-transcode.cpp</code> + clarify ffmpeg usage by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/ggerganov/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/ggerganov">@ggerganov</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4557186287" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3846" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3846/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3846">#3846</a></li> <li>ci : fix path to whisper.h in examples.yml [no ci] by <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/danbev/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/danbev">@danbev</a> in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="4547913309" data-permission-text="Title is private" data-url="https://github.com/ggml-org/whisper.cpp/issues/3842" data-hovercard-type="pull_request" data-hovercard-url="/ggml-org/whisper.cpp/pull/3842/hovercard" href="https://github.com/ggml-org/whisper.cpp/pull/3842">#3842</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ggml-org/whisper.cpp/compare/v1.8.5...v1.8.6"><tt>v1.8.5...v1.8.6</tt></a></p> github-actions[bot]