New Ultrafast Mode Speeds Up GPT-5.6 Sol to 750 Tokens per Second

OpenAI announced a new Ultrafast mode designed to significantly accelerate GPT-5.6 Sol. According to TechCrunch, citing OpenAI's blog, Ultrafast allows GPT-5.6 Sol to operate up to 14 times faster than the standard mode, producing up to 750 output tokens per second. OpenAI notes that previously high-speed AI performance was typically achieved using smaller or specialized models, but now the performance of the most powerful model is combined with faster response generation. The new mode can be used in enterprise tasks requiring rapid information processing, including incident response, customer service, financial market analysis, and e-commerce. Currently, Ultrafast is rolling out in preview and is available only to a select group of customers, with its operation supported through collaboration with chip manufacturer Cerebras. OpenAI has not specified a timeline for full launch, only promising to gradually expand availability as computing capacity grows. Competitors like Anthropic have also released accelerated versions of their models; Claude has a 'fast mode', but it does not provide the speed that OpenAI currently offers.