tag:github.com,2008:https://github.com/ollama/ollama/releases

Release notes from ollama

2026-09-09T21:51:50Z tag:github.com,2008:Repository/658928958/v0.34.0 2026-09-10T06:18:35Z

v0.34.0

<h2>Use Ollama models in ChatGPT Desktop</h2> <p>Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open models. Setup is available from the Ollama app on MacOS.</p> <a target="_blank" rel="noopener noreferrer" href="https://private-user-images.githubusercontent.com/29360864/649245808-e10e299d-c11f-447d-9234-afa855824efe.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.DRswlYEGRLjO9XONxNK7zRyV159l5l2djrZbeDZvt8w"><img width="1374" height="1300" alt="CleanShot 2026-09-08 at 11 04 07 AM@2x" src="https://private-user-images.githubusercontent.com/29360864/649245808-e10e299d-c11f-447d-9234-afa855824efe.png?jwt=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.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.DRswlYEGRLjO9XONxNK7zRyV159l5l2djrZbeDZvt8w" content-type-secured-asset="image/png" style="max-width: 100%; height: auto; max-height: 1300px;"></a> <p>This release also improves structured output performance on Apple Silicon, adds support for OpenAI-compatible client tool search and response compaction.</p> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ollama/ollama/compare/v0.33.3...v0.34.0"><tt>v0.33.3...v0.34.0</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/658928958/v0.34.0-rc5 2026-09-09T21:51:50Z

v0.34.0-rc5

<p>openai: support standalone named function outputs (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5404430682" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18348" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18348/hovercard" href="https://github.com/ollama/ollama/pull/18348">#18348</a>)</p> ParthSareen tag:github.com,2008:Repository/658928958/v0.34.0-rc4 2026-09-09T17:20:48Z

v0.34.0-rc4

<p>proxy: normalize namespaced commands in Full Access (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5393403495" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18331" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18331/hovercard" href="https://github.com/ollama/ollama/pull/18331">#18331</a>)</p> ParthSareen tag:github.com,2008:Repository/658928958/v0.34.0-rc3 2026-09-09T00:09:48Z

v0.34.0-rc3

<p>openai: accept plaintext-labeled Codex agent messages (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5393123059" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18329" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18329/hovercard" href="https://github.com/ollama/ollama/pull/18329">#18329</a>)</p> ParthSareen tag:github.com,2008:Repository/658928958/v0.34.0-rc2 2026-09-08T23:27:20Z

v0.34.0-rc2

<p>openai: finalize responses at the web search limit (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5392730815" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18328" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18328/hovercard" href="https://github.com/ollama/ollama/pull/18328">#18328</a>)</p> ParthSareen tag:github.com,2008:Repository/658928958/v0.34.0-rc1 2026-09-05T00:48:55Z

v0.34.0-rc1

<p>app: harden Codex desktop proxy handling (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5353863911" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18244" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18244/hovercard" href="https://github.com/ollama/ollama/pull/18244">#18244</a>)</p> ParthSareen tag:github.com,2008:Repository/658928958/v0.34.0-rc0 2026-09-04T21:04:48Z

v0.34.0-rc0

<p>app: add Ollama to ChatGPT Desktop (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5352566253" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18236" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18236/hovercard" href="https://github.com/ollama/ollama/pull/18236">#18236</a>)</p> hoyyeva tag:github.com,2008:Repository/658928958/v0.33.3 2026-09-03T17:24:07Z

v0.33.3

<h2>What's Changed</h2> <ul> <li>gemma4 now supports images and audio on MLX engine</li> <li>Report cached prompt tokens</li> <li>Honor GGUF model defined default parameters</li> <li>MLX, MLX-C, llama.cpp update</li> </ul> <h2>New Contributors</h2> <ul> <li><a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url="/users/marcelpetrick/hovercard" data-octo-click="hovercard-link-click" data-octo-dimensions="link_type:self" href="https://github.com/marcelpetrick">@marcelpetrick</a> made their first contribution in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5075035763" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/17579" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/17579/hovercard" href="https://github.com/ollama/ollama/pull/17579">#17579</a></li> </ul> <p><strong>Full Changelog</strong>: <a class="commit-link" href="https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3"><tt>v0.33.2...v0.33.3</tt></a></p> github-actions[bot] tag:github.com,2008:Repository/658928958/v0.33.3-rc2 2026-09-02T22:07:28Z

v0.33.3-rc2: gemma4: image and audio input support

<p>Safetensors gemma4 imports served by the MLX engine now answer image<br> and audio chats. Images run through both vision architectures: the<br> transformer tower (26B, 31B, e-series) and the 12B's encoder-free<br> unified embedder. Audio arrives through the same intake the ollama<br> API already accepts for gemma4 GGUFs — WAV bytes in the images field,<br> OpenAI input_audio parts, and /v1/audio/transcriptions uploads — with<br> the e2b/e4b checkpoints running clips through their conformer audio<br> encoder and the 12b unified checkpoint embedding the raw waveform<br> directly. Clips longer than 30 seconds are split evenly into chunks<br> of at most 30 seconds, cut at pauses, and encoded independently.</p> <p>Each modality serves only checkpoints that carry it: 26B/31B have no<br> audio config and reject audio input, and checkpoints with an<br> unrecognized vision architecture still load as text-only models and<br> reject image requests.</p> <p>The server previously hid the vision and audio capabilities for<br> gemma4 safetensors because the engine served neither. Both<br> suppressions are removed, and existing imports start advertising the<br> capabilities without re-importing since import already records them.</p> jessegross tag:github.com,2008:Repository/658928958/v0.33.3-rc1 2026-09-02T16:46:04Z

v0.33.3-rc1

<p>llama.cpp: version bump b10760 (<a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5326406669" data-permission-text="Title is private" data-url="https://github.com/ollama/ollama/issues/18199" data-hovercard-type="pull_request" data-hovercard-url="/ollama/ollama/pull/18199/hovercard" href="https://github.com/ollama/ollama/pull/18199">#18199</a>)</p> dhiltgen