●Change8

v0.10.0

Breaking Changes
📦 ollama
⚠ 1 breaking✨ 5 features🐛 3 fixes🔧 7 symbols

Summary

Ollama v0.10.0 introduces a new desktop app, significant performance optimizations for gemma3n and multi-GPU setups, and critical fixes for tool calling and API image support.

⚠️ Breaking Changes

  • Parallel request processing now defaults to 1. This may impact throughput for concurrent users unless manually reconfigured.

Migration Steps

  1. If your workflow relies on high concurrency, review the parallel request settings as the default has changed to 1.
  2. Download the new app version for macOS or Windows from the official site.

✨ New Features

  • Added context length display to the 'ollama ps' command.
  • Improved performance for gemma3n models by 2-3x.
  • Improved multi-GPU performance by 10-30%.
  • OpenAI-compatible API now supports WebP images.
  • Enhanced error display in 'ollama run' for better user feedback.

🐛 Bug Fixes

  • Fixed tool calling issues with granite3.3 and mistral-nemo models.
  • Fixed tool calling logic error where overlapping tool names (e.g., 'add' vs 'get_address') caused failures.
  • Fixed an issue where 'ollama show' would incorrectly report an error.

🔧 Affected Symbols

ollama psollama showollama rungemma3ngranite3.3mistral-nemoOpenAI-compatible API