Tag
The 5.6 Sol model is coming to Cerebras hardware in July, offering inference at 750 tokens per second.