Ollama: MLX Upgrades and OpenAI Compatibility Fixes
Three merged MLX changes improve long-context scaling, startup device checks, and new model support, while open pull requests tighten OpenAI API compatibility and template rendering.
Duration: PT2M20S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-11-05T14:01:41Z
- Audio duration: PT2M20S
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 Thursday, November fifth, twenty twenty-six, and this is your Ollama developer briefing.
The lead story is the MLX backend getting more reliable and more capable, with three merges landing together.
First, YaRN scaling support for Qwen 3 and Qwen 3 point 5 in pull request nineteen-two-thirty-one means explicit rope scaling metadata is now honored for long context, including vision and partial rotation cases, with new reference tests for prefill and decode. Second, pull request nineteen-five-twenty-one makes the…
The second theme is OpenAI compatibility hardening. Long-lived agent threads were hitting a real wedge: compaction could keep a prior summary output while dropping its matching call, causing the next compaction to fail validation. Pull request nineteen-five-thirty-two keeps those pairs together. A related fix in…
Finally, a cluster of small correctness fixes: tool function definitions now render as valid JSON instead of Go struct text in nineteen-five-thirty-four, legacy templates preserve helpers and definitions in nineteen-five-twenty-six, and parameter formatting returns a validation error instead of panicking on missing…
Also merged: the llama dot C P P…
Nearby episodes from Ollama
- Weekly Recap - MLX Model Expansion and Launch Polish
- Ripple on MLX and Validation Hardening
- MLX Memory Cuts and Tool-Call Fixes
- Launcher Return, API Fixes and Inference Reliability
- Launch Hub Expands, Tokenizer APIs Proposed
- Reliability Fixes and Rockchip NPU Support
- Inference Correctness and Installer Reliability
- Weekly Recap - Sampler Fidelity, Faster Drafting, and Launcher Fixes