Can't wait for 8x22B to make it to Groq! Having an LLM at near GPT-4 performance with Groq speed would be incredible, especially for real-time voice chat.

I also assume groq is 10-15x faster