GPT-6.1 Sol: near-Astra quality at one fifth the price

The post announcing GPT-6.1 Sol makes two promises, and only one applied to Sol on launch day. In a September 29 post, Tibo of OpenAI introduced 6.1 Sol as offering near-Astra intelligence at one fifth of Astra's price, with a 95% discount on cache reads. Astra is the pricier model Sol is being compared with. "Near Astra" is how he describes it, and the post cites no benchmark to back that up. A cache read happens when a request repeats input the model has already processed, such as a long shared prompt or document. That reused part is billed at the discounted rate, so the discount matters most for work that sends the same context over and over.

The speed claim is about a different product. Ultrafast, which the post says makes output 8X faster, was available that day for Astra and only described as coming soon for 6.1 Sol. Anyone picking Sol at launch got the lower price but not the faster tier. OpenAI's companion Ultrafast announcement introduces that tier.