Ollama: Weekly Recap - Decision Models Go Live, MLX Gains, and API Polish
Ollama shipped runnable decision-model workflows alongside major MLX runtime and new-model enablement. Cloud proxy, Responses compatibility, and Windows and server reliability fixes rounded out the week.
Duration: PT3M23S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-10-19T09:09:02Z
- Audio duration: PT3M23S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Hello and welcome to your Ollama recap for October twelfth through the nineteenth.
This week tracked 50 pull request activity items, 30 additional commits this week.
The lead insight: Ollama made decision models runnable, testable, and consistent across backends.
First, decision scoring became a real workflow. You can now use ollama run directly with decision models like Nimble and Clef, asking yes-or-no questions with ask, choice, or level scoring, reading state from standard input or arguments, covered in PR eighteen nine sixty-two. A follow-up, PR nineteen oh forty-one,…
Second, the MLX runtime and model coverage expanded. Version bumps in PRs eighteen nine fifty-six and eighteen nine ninety-three picked up upstream Metal fixes for M one and M two systems. PR eighteen six fifty-seven ported Metal custom kernels for Mamba scan and gated operations to CUDA, ending graph fallback. PR…
Third, cloud, compatibility, and reliability polish. For the Open A I Responses interface, PR nineteen oh oh four clarifies sign-in errors, PR nineteen oh oh seven preserves upstream Chat G P T errors, PR nineteen oh ten tolerates reasoning effort on non-thinking models, and PR eighteen nine eleven…
Nearby episodes from Ollama
- Remote Cancel Fixes and Wider Model Support
- MLX Compatibility Push and Responses Fixes
- Rendering Transparency and Response Fidelity Fixes
- Proxy Transfers and API Resilience
- Model Compatibility Catch-Up and Windows Reliability
- Responses Compatibility and Clearer Errors
- Decision Models Go Usable, MLX Speeds Up on CUDA
- Decision Scoring Matures as Inference Hardens