Groq

Very fast inference for open models.

freemium

Best for: Latency-sensitive apps