OpenAI cut the price of its best model by a third on output — and put a three-month clock on it
OpenAI announced on 21 August that GPT-5.6 Sol, its frontier model, drops from 5 to 4 dollars per million input tokens and from 30 to 20 dollars per million output tokens, with cached input falling from 50 to 40 cents. Output is where most of the bill comes from, so for anything that writes long answers the effective saving is well past the headline 20 percent. The cut applies to the API and to eligible Codex and ChatGPT Work credit plans; Pro, Plus and Business subscriptions are unchanged. It runs until 21 November rather than permanently. OpenAI cut prices on the smaller Terra and Luna models on 30 July, and those reductions were permanent. Reuters reports the move as a response to pressure from Anthropic and from Chinese models.
Why it mattersA temporary cut on the flagship, aimed only at customers who pay by the token, reads as a defensive move rather than an efficiency dividend: if the model had simply become cheaper to run, the price would have stayed down. For anyone with a product on the API it is real money for one quarter and a decision to make in November. For everyone else it is a market signal — frontier capability is no longer being priced as a scarce good.
▶ Related video: OpenAI slashes prices amid strong competition
Read in the app — free, in 9 languages
Related stories
Waymo has quietly built its own chip, and it is already riding around in the cars2026-08-22The man who shipped Google's first seven generations of TPU has gone to Anthropic
2026-08-22Anthropic expects to raise as much as SpaceX did — the largest listing in history — and could file publicly this month
2026-08-21Nvidia is paying $6 billion for the machine that builds a rival's models — and hiring 109 of the people who ran it
2026-08-21The doctors and lawyers who grade AI answers are now a half-billion-dollar business — and this one grew fivefold in eight months
2026-08-21