The new Ultrafast mode delivers up to 750 tokens per second, a significant leap that challenges the industry standard of sacrificing model intelligence for raw speed. Previously, developers seeking rapid interactions were forced to rely on smaller, specialized models. By optimizing the performance of its flagship GPT 5.6 Sol, OpenAI intends to facilitate high-stakes workflows such as financial market analysis, incident response, and customer support where latency can be a critical barrier.
This performance increase is powered by a strategic partnership with chipmaker Cerebras. While the technology is currently restricted to a limited preview group, the company plans to broaden access as compute capacity grows. The move places OpenAI in direct competition with rivals like Anthropic, whose own fast-mode iterations for Claude have set a baseline for speed that this latest development aims to surpass.
Comments (0)
No comments yet. Be the first!