Ollama: Tool Reliability and Mac Inference Tuning
Tool-calling correctness dominates the activity, with merged deferred tool support and fixes to preserve schema details and handle edge-case values. Mac runner efficiency, long-context stability, and desktop usability round out the day.
Duration: PT2M21S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-11-13T14:01:22Z
- Audio duration: PT2M21S
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, November 13th, 2026, and this is your Ollama developer briefing.
The through-line today is correctness under the hood: tool calls that survive real-world schemas, and inference that stays fast without degrading.
First, tool calling gets more robust. Merged support for deferred tool loading in Claude Code, tracked in P R 19578, aligns Ollama with how that client stages tools. Two open proposals carry that further. P R 19686 would preserve unmodeled schema keywords — things like default and minimum — that are currently…
Second, Apple Silicon performance and long-context stability. P R 19677 lets a new request reuse a compatible saved prefix even from a busy slot, cutting repeated prompt processing while keeping each request independent. P R 19678 makes saved history adapt to memory pressure instead of a fixed 8 gigabyte allowance,…
Finally, day-to-day usability. An open proposal in P R 19683 would add Explore and Saved Models to Ollama Desktop, so you can browse the public library and shortlist models without leaving the app. Merged fixes help now: bare ollama prints help in non-interactive sessions instead of trying to launch the interface,…
What's next:…
Nearby episodes from Ollama
- Faster Concurrent Inference and Unified Desktop Launch
- MLX Goes Parallel
- VRAM Fixes and Launch Polish
- Stricter Model Imports and CLI Output Fixes
- 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