OpenAI introduces ‘Ultrafast,’ a new mode that makes GPT-5.6 Sol work at 14x the speed
TechCrunch Lucas Ropek ● Covered by 2 sources
OpenAI launched Ultrafast for GPT-5.6 Sol, claiming 14x faster output. It’s in preview for a small group now, with a Cerebras partnership behind it.
Based on reporting by TechCrunch, Lucas Ropek — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI has added a new mode called Ultrafast to GPT-5.6 Sol, its latest and most powerful model, and the pitch is simple: make it move much faster. The company says the mode can run at 14 times the speed of standard processing and produce up to 750 output tokens per second.
That matters because speed has often forced a tradeoff. OpenAI said that if you wanted real-time performance before, you usually had to settle for a smaller or more specialized model. Ultrafast is supposed to change that by pushing “more useful work per second” from a bigger model instead.
The company is aiming this at business tasks where latency matters: incident response, customer service and support, financial market analysis, and e-commerce. And while that’s a familiar AI sales pitch, the numbers here are eye-catching enough to make the claim hard to ignore.
Ultrafast is only in preview for now, and access is limited to a small group of customers. OpenAI says it will widen access as capacity grows, and says the feature is powered by its partnership with Cerebras, the chipmaker behind the speed boost.
My take — AI-written commentary, not fact-checked reporting
This is the real race now: not just smarter models, but models that can spit answers out before people lose interest. OpenAI is betting that speed sells, and it probably does, especially in support and trading. The awkward part is that the industry keeps dressing up throughput as progress, which is fine until everyone starts confusing faster with better.
Read more about this at: TechCrunch