Open-weight models for realtime voice and agent workloads. Click a card for specs, sources, and how to call the API.
1.7B · Streaming speech-to-text
