Ollama released v0.40.11 on October 24, 2026. The release promotes the v0.40.11 release candidate to stable and broadens hardware availability for the features introduced in v0.40.9: speech generation and the d3 decision models are now described as available on macOS, not only Apple Silicon. Read the release notes.
What is now generally available
- Speech:
ollama runplays text or a chat model's reply aloud as the audio is generated, and applications can stream audio through the OpenAI-compatible/v1/audio/speechendpoint, for example withollama run phonon "Hello from Ollama." - d3 decision models: the vLLM Semantic Router family (2B, 4B, 9B, 27B) classifies input, answers yes-or-no questions or scores options with probabilities from text, structured state or images.
The changelog against v0.40.10 lists these availability changes plus a first community contribution. For the original feature announcements and the Nemotron Bolt E4B addition, see our reports on v0.40.9 and v0.40.10.