Ollama: Truth in Reporting
Four fixes landed today, all centered on one theme: making Ollama's status reporting match reality — accurate capability listings, accurate memory numbers, accurate error codes, and safer parsing of untrusted input.
Duration: PT2M21S
Episode overview
This episode is a short developer briefing from Ollama.
It explains recent repository work in plain language.
- Show: Ollama
- Published: 2026-07-05T13:01:01Z
- 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 day, and welcome to Ollama for July 5th, 2026.
The throughline across today's activity is accuracy — making sure what Ollama reports to users and clients actually reflects what's happening under the hood.
Start with capability reporting. PR 17038 fixes a mismatch where the "tags" endpoint failed to show tool support for models like deepseek-r1, even though the "show" endpoint got it right. The root cause was in how the model list summary checks a template's variable list — it was missing a case that the other…
Memory reporting gets a similar correction. PR 17037, from sean-kim05, found that sliding-window attention models — the gemma2 and gemma3 family — were under-reported in VRAM accounting by a full key-value cache once context length grew. On a sixteen gigabyte card running gemma3, actual usage at 128K context was…
Error reporting rounds out the theme. PR 17039, also from sean-kim05, found the legacy embeddings handler returning a 500 server error for what's actually a client mistake — input longer than the model's context. The newer embed endpoint already returns a 400 for the same case. Now both agree.
And underpinning all of this: PR 17036 closes three separate panics —…
Nearby episodes from Ollama
- MLX Create Pipeline Rewrite Lands
- Agent Harness Lands, Hardware Support Gets a Cleanup
- Gemma 4 Support and Platform Improvements
- Weekly Recap - MLX Performance & Path Handling
- Memory Management and Multimodal Parsing Fixes
- GPU Offloading and Tool Call Fixes
- Performance Optimizations and Model Handling Improvements
- Infrastructure Updates and Platform Fixes