Ultra-low-latency inference cloud powered by custom LPU hardware.
Groq runs open-source LLMs on its custom Language Processing Unit (LPU) hardware, delivering exceptionally low latency and high tokens-per-second throughput. GroqCloud offers an OpenAI-compatible API with usage-based per-token pricing and a free developer tier with rate limits.
Free developer tier with rate limits; usage-based per-token pricing on paid tiers; enterprise/on-prem available.