Ollama: Silent Failure Fixes and Mac Expansion
Today's activity centers on stopping silent data loss in compatibility APIs and tool calls, alongside expanded Apple hardware support. Scheduler, memory reporting, and launch fixes improve reliability for single-model hosts and Mac-local workflows.
Duration: PT2M33S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-10-22T13:01:09Z
- Audio duration: PT2M33S
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 October 22nd, 2026, and this is your Ollama developer briefing.
The throughline today is preventing silent data loss, with a second push to make Mac-local models faster and easier to run.
First, compatibility layers are getting stricter about what they preserve. In PR 19188, Anthropic streaming no longer swallows runner failures after the stream starts — it now emits a proper error event instead of leaving clients with partial text and no cause. Related Open A I work tightens how developer…
Second, Apple hardware support is expanding. PR 19184 adds a runner for Apple Foundation Models on the new mac O S, exposing it as apple-foundation-model for run, chat and Open A I compatible requests with no download, since the system manages the weights. On the M L X side, PR 19197 adds concurrent generation with…
Third, scheduling and launch behavior is more accurate. PR 19190 fixes an eviction stall when only one model slot is allowed, where a keep-alive could cancel a pending eviction. PR 19205 fixes split log lines that caused under-reported memory and layer counts, and launch work now reads allocated context before…
Merged updates include the Flint model support in commit…
Nearby episodes from Ollama
- Weekly Recap - Phonon Voice Goes GA and MLX Speed Gains
- Gemma 4 Reliability and Runner Parity
- Tool Call Fixes and Phonon Voice Update
- Phonon Voice Goes Stable and API Failures Get Fixed
- API Compatibility and Runtime Reliability
- Remote Cancel Fixes and Wider Model Support
- MLX Compatibility Push and Responses Fixes
- Weekly Recap - Decision Models Go Live, MLX Gains, and API Polish