Ollama has released version 0.31.2, introducing several improvements and fixes. Key updates include enabling flash attention on older NVIDIA GPUs (compute capability 6.x) and allowing iGPUs to offload vision models by fitting them into available memory. The release also addresses issues with structured output for thinking models, hardens GGUF model creation, and disables telemetry by default for `ollama launch` with Claude Code. Additionally, a fix has been implemented for loading models on paths containing non-UTF-8 characters. AI
IMPACT Improves performance and memory handling for local LLM deployments, particularly on older hardware.
RANK_REASON This is a software release for a tool that facilitates running LLMs locally, not a frontier model release.
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
