OpenAI has stopped generating revenue through its most expensive AI models, according to a report from Chinese business outlet 36 Kr, surfaced via Google News.
That single line is, at the moment, the whole of the available reporting. The item circulating carries the headline and the outlet's name, but no accompanying figures, executive quotes, timeline, or list of which specific models are affected. It is worth being blunt about that gap rather than papering over it: anything beyond the claim itself — why the change was made, whether it is permanent, how much money is involved — is not established by the source at hand.
Still, the claim is a notable one if it holds up. An AI company's most expensive models are usually its most capable ones, the tier sold to businesses and power users at premium prices. Revenue from that tier is what helps offset the enormous cost of training and running frontier systems. A company choosing not to charge for it — or no longer able to — cuts against the standard assumption that the smartest models are also the most lucrative.
Readers should treat this as an early, thinly sourced signal rather than a settled fact, and watch for confirmation from OpenAI or from outlets with independent reporting.
It matters because how AI companies make money from their best models shapes what those models cost everyone else — and whether the industry's spending can be sustained.