Code · Groq
Groq (GroqCloud)
Custom-chip inference, 500+ tokens/sec
[ Overview ]
GroqCloud serves open-weight models on custom LPU chips, delivering some of the fastest inference speeds in the industry at low per-token cost.
[ Pros ]
- Exceptional inference speed (~10x GPU hosts)
- Low per-token pricing
- Simple OpenAI-compatible API
[ Cons ]
- Limited to models Groq has optimized for LPUs
- No custom fine-tuning
[ Tags ]
inferenceLPUlow-latencyopen models