v0.31.2

knadh/listmonkv0.31.2Jul 6, 2026by github-actions[bot]

AI Summary

This release enhances hardware compatibility by enabling flash attention on older NVIDIA GPUs and improves memory handling for iGPUs when offloading vision models. It also addresses stability issues, including fixes for structured outputs and non-UTF-8 model paths, alongside updates to underlying inference engines.

Key Highlights

Enabled flash attention on older NVIDIA GPUs (compute capability 6.x)
iGPU can now offload vision models with padding to fit available memory
Fixed structured output for thinking models when thinking is disabled
Hardened GGUF model creation
Updated the MLX and llama.cpp engines

New Features

Support for flash attention on older NVIDIA GPUs
iGPU offloading for vision models with padding
Disabled telemetry by default in `ollama launch` for Claude Code

Full Release Notes

## What's Changed

* Enabled flash attention on older NVIDIA GPUs (compute capability 6.x)
* iGPU can now offload vision models with padding to fit available memory
* Fixed structured output for thinking models when thinking is disabled
* Hardened GGUF model creation
* `ollama launch` for Claude Code now disables telemetry by default
* Fixed loading models on paths with non-UTF-8 characters
* Updated the MLX and llama.cpp engines

## New Contributors
* @kevinpark1217 made their first contribution in https://github.com/ollama/ollama/pull/16949

**Full Changelog**: https://github.com/ollama/ollama/compare/v0.31.1...v0.31.2