Instruction-tuned mixture-of-experts model for high-volume work: classification, extraction, summarisation, routing and short chat. Chosen when throughput and cost matter more than depth of reasoning. OpenAI-compatible; streaming supported.

Payment Options