Ollama v0.32.6
What's Changed Qwen3.5 is faster on Apple GPUs: the MLX engine now uses the model's MTP head for speculative decoding automatically /v1/chat/completions streaming now matches OpenAI's wire format: role only on the first chunk, finishreason on its own…
Source ollama/ollamaPublished 10d ago · Aug 5, 2026Posted on Bluesky
What we hold
- Repo
- ollama/ollama
- Version
- v0.32.6
Getting releases like this one by email is a separate problem from reading them here. GitHub’s custom watch, a repository’s releases.atom feed, Dependabot and NewReleases.io, compared with sources: how to get an email when a GitHub repo publishes a release.
More like this
Next.js v16.3.1
What's Changed [16.x] Turbopack: don't strip async-module runtime from shared runtime chunks by @lukesandberg in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5063455839" data-permission-text="Title is private"…
workerd v1.20260813.1
What's Changed Fix segfault in JSG unwrap functions by @logan-gatlin in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5121651787" data-permission-text="Title is private"…
Next.js v15.5.23
What's Changed [15.x] Port ReplyServer traversal guards to FlightClient @eps1lon in <a class="issue-link js-issue-link" data-error-text="Failed to load title" data-id="5024206928" data-permission-text="Title is private"…