Update: Ollama 0.40.6 Is Out, and 0.40.7-rc0 Brings Responses Effort and PPLX Decider

Update to my earlier post, Ollama 0.40.5: Nemotron 3.5 Super and Bolt for Local Agents. Ollama has moved on twice since then.

Ollama 0.40.6 (stable, Oct 16)

• Windows: prevents orphaned runner processes.
• GGUF: adds the q2_0 tensor and file types.
• Launch clef in Ollama’s cloud.
• --ask mode now uses identifiers for question keys.

Ollama 0.40.7-rc0 (pre-release, Oct 17)

• OpenAI-compatible API: Responses “effort” is now allowed on non-thinking models.
• llama-server can infer the draft speculative-decoding type on its own.
• MLX: adds PPLX Decider v1.2 support.

My take

0.40.6 is a safe update, especially on Windows. The 0.40.7 release candidate is mostly plumbing for people using Ollama behind OpenAI-style clients. I’d wait for the final release unless you need one of those fixes.

References & Further Reading

• Ollama v0.40.7-rc0 release (GitHub)
• Ollama v0.40.6 release (GitHub)
• Ollama 0.40.5: Nemotron 3.5 Super and Bolt for Local Agents (EtherNinja)