News · 2026-10-25
Ollama v0.40.11 adds local speech generation on macOS with an OpenAI-compatible audio endpoint
Ollama v0.40.11, published 24 October 2026, adds speech generation on macOS: `ollama run` plays a text prompt or a chat model's reply aloud as the audio is generated, and apps can stream speech through an OpenAI-compatible /v1/audio/speech endpoint. The same release adds d3, a family of decision models (2B, 4B, 9B and 27B sizes) from vLLM Semantic Router that classify text, structured state or images, answer yes/no questions, or score options with probabilities.
What it means for buyers
Speech output closes the last big modality gap for a fully local voice loop on the Mac: local speech recognition plus local speech generation can now run with no cloud call at all, putting a basic voice-agent stack in reach of people already running local models — a cost contrast with per-minute voice-agent APIs like Vapi and Bland. The d3 router models are a cheap way to gate which requests reach an expensive model. macOS only for now; the release notes list no Windows or Linux speech support.
Where to check it
- Ollama profile: dated claims, sourced pricing
Sources
- Release v0.40.11 (Ollama GitHub release notes) — speech on macOS, /v1/audio/speech endpoint, d3 decision models, published 2026-10-24, checked 2026-10-25
- Speech capability docs (Ollama docs) — speech capability reference linked from the release notes, checked 2026-10-25