Ollama: Model Coverage and Tool-Call Reliability
New model support for Nemotron 3 and Gemma 4 image input expanded Ollama's MLX backend, while a cluster of parser fixes targeted tool-call reliability and error clarity for developers building agentic workflows.
Duration: PT2M55S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-08-11T13:01:27Z
- Audio duration: PT2M55S
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 August 11th, and here's what moved in Ollama's codebase.
The headline is model expansion paired with a hard look at tool-call reliability. On the model side, Daniel Hiltgen's PR 17060 brings Nemotron 3 Nano Omni to MLX, including recurrent Mamba pieces, mixture-of-experts routing, and a self-draft speculator that skips the need for a separate draft model during…
The second theme is tool-call and parser hardening. Three separate PRs converge here. PR 17664, merged by Hiltgen, fixes a parsing bug where boundary tokens leak into invoke names during tool calls, replacing a fragile strip-based fix with a shared name-scanning approach. PR 17651 from xoloki wraps tool-call parse…
Smaller but practical: PR 17649 adds per-file hashing progress to "ollama create," addressing a real user complaint about a silent 30-to-60-second phase during large model imports. PR 17654 fixes a Windows-on-Arm build flag that was shipping CPU binaries without dot-product instructions. And PR 17657 fixes a…
On docs, VS Code guidance was updated twice — a minimum version bump to 1.127 for reliable request cancellation, and new guidance on context length mismatches between VS Code and Ollama's…
Wha…
Nearby episodes from Ollama
- Qwen Gets Serious About Coding Agents
- Launch Ecosystem Expands, Parser Edge Cases Get Squashed
- Backup Collisions and a Security Fix Converge
- Fixing the Defaults That Silently Hurt You
- Vision Support Lands in the MLX Runner
- Weekly Recap - Vision Runners and Robustness Cleanup
- Edge Case Cleanup Across the Stack
- Speculative Decoding Gets a New Draft Architecture