OllamaProduct update
Ollama · v0.40.14
Ollama v0.40.14 makes ollama launch claude honor each model's output token limit instead of the 128k default, and OpenAI-compatible endpoints keep a model's packaged sampler defaults when clients omit temperature or top_p. The app can retry ChatGPT disconnects, launch codex checks ChatGPT routing first, compressed cloud web-search responses decode correctly, and top-k sampling on MLX is faster for large vocabularies.
`ollama launch claude` now applies the model's output token limit for Claude Code, so cloud models no longer run into the 128k default cap.
Saved . A shortened excerpt from GitHub Releases; the original publisher retains authorship.