Ollama released version 0.40.10 on 23 Oct 2026. The update adds support for NVIDIA’s Nemotron 3.5 Bolt E4B model. The model is built on the Gemma‑4 architecture, works locally, and can handle image input, tool use, and long contexts up to 128 000 tokens. It also includes a DFlash draft for speculative decoding, which speeds up generation. Users can start it with ‘ollama run nemotron-3.5-bolt:e4b’.
Why it matters
Local developers can now run a fast vision‑language model on their own machines without cloud calls.
