Ollama v0.40.7 ships pplx-decider, a Perplexity 27B decision model pulled through the client
Ollama published v0.40.7 on Oct 18, 2026. The release adds pplx decider, a 27B decision model from Perplexity fine tuned from Qwen3.8 27B and available via oll…
Published on MyPrivateClaw
Oct 21, 2026, 4:19 PM UTC
Coverage date
Oct 21, 2026
Last updated
Oct 21, 2026, 4:19 PM UTC
News summary
Ollama published its v0.40.7 release on Oct 18, 2026, a new inference runtime/model server update for the Ollama client and API. The release adds pplx decider, a 27B decision model from Perplexity, fine tuned from Qwen3.8 27B, available via ollama pull. The version also no longer fails continuing an OpenAI compatible Responses conversation when switching between models with and without thinking, or when the history contains another provider's compaction items. Separately, v0.40.7 now cancels requests routed to remote models on ollama.com when the client disconnects instead of continuing to run. Why it matters v0.40.7 changes two constraints on ongoing work: what a conversation history may contain, and what happens to a request once the client goes away. A saved OpenAI compatible Responses history can now be continued across models with and without thinking, and across history that conta…