Sources: OpenAI is considering drastically lowering the prices it charges users for tokens in anticipation of similar cuts the startup expects Anthropic to make
The company might lower prices for tokens, the central unit for gauging AI costs, though the discussions are still in flux
Wall Street Journal
Context & Ripple Effects
Related coverage already shows enterprise buyers responding to rising AI costs by routing work to cheaper models, including some from China. That puts OpenAI and Anthropic under pressure not only on model capability but on the unit economics of serving workloads.
The reported pricing discussions also sit alongside later reporting that OpenAI had found a way to cut inference costs substantially, linking any potential price move to the underlying cost of delivering models rather than to a standalone promotion.
First-order effects
OpenAI’s token pricing is reportedly under review, with the company preparing for the possibility that Anthropic makes comparable reductions; no change is final while discussions remain in flux.
Enterprise AI customers and developers would gain stronger leverage in current procurement and usage decisions as two leading suppliers signal potential lower per-token costs.
Second-order effects
Cheaper models already being adopted for cost control become a more direct benchmark for OpenAI and Anthropic: the leading vendors may need to compete more explicitly on price as well as performance.
If lower inference costs enable the cuts, suppliers that can reduce serving costs fastest will have more room to defend margins while matching customer price expectations.
Third-order effects
If repeated across providers, token pricing could become less of a premium differentiator and more a pass-through of inference efficiency, shifting competitive advantage toward low-cost infrastructure and model routing.
The likely result is a more segmented market: premium models retain pricing power only where their performance justifies it, while routine workloads increasingly move to the lowest-cost acceptable option.
The trend: This is one data point in AI-model pricing moving from capability-led premiums toward a cost-efficiency and workload-routing contest.
*OpenAI Price Cuts Could Threaten Margins, Profits Ahead of Potential IPO, Sources Say — WSJ maybe should have rushed that IPO out while nobody was checking their token bills... too late now
OpenAI is considering drastic price cuts to compete with Anthropic. OpenAI cannot under any circumstances afford this pay cut. This is fundamentally insane. — www.wsj.com/tech/ai/open... [image]
Yeah, this isn't a sign of strength, Sam.. when you need to start cutting prices to keep customers, it's not a good look ahead of an IPO. OpenAI may just be the odd man out of the OAI/Anthropic/SpaceX IPO triumvirate [embedded post]
OpenAI and Anthropic may begin a race to the bottom for token prices, but it's not clear who is going to fire the first shot. This might be driven by OpenAI lacking anything Mythos tier and trying to move into the budget sector before Sonnet and Opus 5.
I was just an ignorant humanities grad student at the time but I think that this was what the start of the dot-com bubble looked like www.wsj.com/tech/ai/open...
With companies like SpaceX, Anthropic, and OpenAI potentially heading toward IPOs, investors are already debating which one will be the biggest winner. …
If OpenAI does drop token pricing, this is likely because they've heard from customers they can't adopt AI at volume at the current pricing. Margin is high now for served tokens. They could cut prices by like 60% and still be profitable imo
Is an AI price war coming? — OpenAI is reportedly considering drastically lowering the prices it charges users for tokens (https://lnkd.in/eN4ZKjXR). …
The competition between OpenAI and Anthropic has become a blood sport. What do they do? Lower their prices to attract customers, or raise them to lose less money? …