OpenAI is considering drastic cuts to the token prices it charges developers and enterprises, according to a new report. This move anticipates a potential price war with rival Anthropic as both companies, neither yet profitable, have confidentially filed for IPOs. The pressure is intensified by a sharp decline in ChatGPT’s web traffic share and the explosive growth of Anthropic’s revenue, driven by its coding tool. Meanwhile, open-source inference providers are offering powerful Chinese models at a fraction of the cost, challenging the financial models of Western AI labs.
OpenAI is weighing significant reductions in its token pricing, per a recent report. The discussions come as both OpenAI and Anthropic have filed for IPOs this month while operating at a loss. “I think we’ll have a lot of ways we can help people get more value for less spend,” Sam Altman stated recently.
OpenAI reported a -122% adjusted operating margin in the first quarter of 2026. Anthropic‘s annualized run rate surged to $47 billion by May 2026, a 422% increase driven largely by its Claude Code product. ChatGPT’s share of global generative AI web traffic fell from 77.6% to 53.7% over a similar period.
Enterprise AI spending is skyrocketing, a practice dubbed “tokenmaxxing.” Palantir CEO Alex Karp compared the trend to a porn addiction. JP Morgan analysts published a note titled “AI Bills Are Out of Control.”
The flat-fee consumer model was a loss-leader, as explained by Tommy Shaughnessy of Delphi Ventures in a widely shared analysis. Real enterprise deployments use metered API pricing, causing costs to balloon. This exposes the companies to customer blowback as they contemplate a price war.
Open-source inference providers now serve models like DeepSeek V4 at roughly one-thirteenth the price of closed alternatives. Shaughnessy noted that Chinese labs open-sourcing frontier models gives inference providers their biggest cost for free. This dynamic continuously pushes the floor on intelligence pricing toward zero.
