Cerebras powers ultrafast mode for OpenAI's GPT-5.6 Sol

Cerebras wafer-scale architecture delivers 750 tok/s, 14x faster than GPU inference, directly challenging Nvidia's inference dominance. Could reshape enterprise deployment decisions for real-time agentic workflows. Raises question of scalability beyond preview. New details: Invite-only preview, pricing not yet disclosed, community skepticism about quality parity. Partnership signals wafer-scale inference gaining traction.