Ollama: Model Compatibility Catch-Up and Windows Reliability
Ollama's runtime caught up to new model formats with a llama.cpp bump, Q2_0 support, and Nemotron MLX fixes, alongside a Windows fix for orphaned runners and agent prompt corrections. Docs and community listings made up the rest of the churn.
Duration: PT2M9S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-10-16T13:01:09Z
- Audio duration: PT2M9S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good morning, it's Friday, October 16th, 2026, and this is the Ollama daily briefing.
The lead signal is compatibility: the core runtime is racing to load models the ecosystem is already shipping.
First, model format support. Pull request 19032 bumps llama dot cpp to build 11669, and pull request 18967 adds the Q2_0 tensor and file types to Ollama's GGUF reader. Without that second change, create and load fail with an unsupported tensor error. The practical payoff is support for Prism ML's Bonsai 2 models in…
Second, reliability and agent correctness. Pull request 18500 assigns Windows runner processes to a system job object set to terminate on close, so llama server and MLX subprocesses are cleaned up even if Ollama is force-closed. That fixes orphaned processes for command-line serve users; desktop shutdown was already…
What's next: if you ship Q2_0 models or run serve on Windows, verify startup and shutdown behavior on the latest build, and watch the truncation and OpenAI logging proposals for changes to prompt construction and API observability.
That's it for today.
Nearby episodes from Ollama
- Proxy Transfers and API Resilience
- Responses Compatibility and Clearer Errors
- Decision Models Go Usable, MLX Speeds Up on CUDA
- Decision Scoring Matures as Inference Hardens
- Native Decision Models and Precision Fixes
- Weekly Recap - Decision Models Go Native, Embeddings Go Multimodal
- GPU Discovery Fallback and Docs Cleanup
- OpenAI Parity and Reasoning Fixes