This release includes breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Topics
+8 more
ReleasePort's take
Light signalOllama v0.32.1 resolves a recurrent MLX model cache leak that could cause memory usage to grow across requests and enhances cache snapshot performance.
Why it matters: Fixes a bug (severity 40) that caused unbounded memory growth in MLX caches; improves stability for all deployments using Ollama's MLX features.
Summary
AI summaryFixed recurrent MLX cache leak and improved memory usage across requests.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Low |
Improved Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations Improved Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Feature | Low |
MLX text model loading now respects `OLLAMA_LOAD_TIMEOUT` environment variable MLX text model loading now respects `OLLAMA_LOAD_TIMEOUT` environment variable Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Feature | Low |
Agent web search and fetch now prompt users to run `ollama signin` when authentication is required Agent web search and fetch now prompt users to run `ollama signin` when authentication is required Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Feature | Low |
Interactive agent now receives the current working directory for better project context Interactive agent now receives the current working directory for better project context Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Bugfix | Medium |
Fixed recurrent MLX model cache leak that could increase memory use across requests and improved cache snapshot performance Fixed recurrent MLX model cache leak that could increase memory use across requests and improved cache snapshot performance Source: llm_adapter@2026-07-16 Confidence: high |
— |
| Bugfix | Medium |
Fixed `ollama launch` so selecting **Pick another model** for a deprecated model passed with `--model` correctly opens the model picker Fixed `ollama launch` so selecting **Pick another model** for a deprecated model passed with `--model` correctly opens the model picker Source: llm_adapter@2026-07-16 Confidence: high |
— |
Full changelog
What's Changed
- Improved Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations
- Fixed a recurrent MLX model cache leak that could increase memory use across requests, and improved cache snapshot performance
- MLX text model loading now respects
OLLAMA_LOAD_TIMEOUT - Agent web search and fetch now tell users to run
ollama signinwhen authentication is required - The interactive agent now receives the current working directory for better project context
- Fixed
ollama launchso choosing Pick another model for a deprecated model passed with--modelopens the model picker - Updated VS Code setup documentation for the official Ollama extension
Full Changelog: https://github.com/ollama/ollama/compare/v0.32.0...v0.32.1-rc0
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About ollama
Get up and running with Kimi-K2.5, GLM-5, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Beta — feedback welcome: [email protected]