Ollama: Weekly Recap - New Model Support & Concurrency Hardening
This week's development centered on landing full support for the new Laguna model family across Metal and MLX backends, alongside a cluster of fixes for scheduler race conditions and tool-calling reliability. The team also began rolling out an AI-assisted PR review agent for maintainers.
Duration: PT3M29S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-07-27T09:08:55Z
- Audio duration: PT3M29S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good week to you, and welcome to the Ollama Weekly Recap for July 20th through 27th. Fifty pull request activity items and thirty additional commits crossed the repository this week.
The headline story is Laguna. Daniel Hiltgen led a multi-PR push to bring the new Laguna model family fully into Ollama: chat template and parser support with a Metal inference fix in PR 17291, alignment with upstream llama-cpp to shed Ollama's local implementation in PR 17335, and full MLX support for Apple Silicon…
The second theme is scheduler and concurrency reliability. PR 17376 fixed a data race in the "ps" handler where the scheduler's loaded-model map was read without proper locking. PR 17416 stops an expiring model runner from being resurrected by a concurrent request. And an open PR, 17318, targets cross-talk in the…
Third, tool-calling correctness got sustained attention. PR 17250 finalizes incomplete tool calls from GLM models instead of silently returning empty responses. PR 17324 stops content from being dropped after bare JSON tool calls. PR 17354 accepts malformed Qwen tool calls missing their wrapper tags. And PR 17407…
Rounding out the week, the agent TUI continued maturing under Parth…
N…
Nearby episodes from Ollama
- Closing the Gaps Between Local Config and Cloud Compatibility
- MLX Performance Push and a Scheduler Race Fix
- A Server Reliability Sweep
- Tool Calling Gets a Reliability Pass
- Scheduler Race Conditions Take Center Stage
- Command Line Consolidation and Hardware Fixes
- Thinking Models Leak Control Tokens
- Weekly Recap - Hardening the Runtime, One Edge Case at a Time