Chat Models
Groq
LLaMA 3.3 70B Versatile
Proven versatile model with excellent reasoning at roughly 280 tokens per second.
Reasoning280 tok/s
Same provider
More from Groq
Groq
GPT-OSS 120B
Open-source GPT running on Groq LPU at roughly 500 tokens per second.
Groq
GPT-OSS 20B
Ultra-fast lightweight GPT running at roughly 1000 tokens per second on Groq LPU.
Groq
Compound Mini
Lightweight compound agent with tool use at roughly 450 tokens per second.
Groq
Compound
Groq compound AI system with tool use and web search.