OpenAI weighs drastic token price cuts to counter Anthropic

Per WSJ-sourced reporting, OpenAI is weighing meaningful cuts to ChatGPT and API token prices in anticipation of an intensifying 'war for users' with Anthropic. The timing — days after Fable 5's capability leap — led developers to read the contemplated cuts as a concession that Anthropic, at least momentarily, won on raw capability, leaving OpenAI to compete on price and distribution.

The economics are the story: pricing is becoming the primary lever as frontier capabilities converge. DeepSeek's permanent 75% V4 Pro cut, Google's AI Plus price reduction, and now OpenAI's contemplated cuts point to margin compression across the board. Microsoft's Satya Nadella simultaneously warned employees against 'tokenmaxxing' — 'don't use frontier models for non-frontier problems' — capturing the cost-discipline mood as the bills come due.

This sits inside OpenAI's larger maneuvering: a confidential IPO filing reportedly targeting as early as September 2026, the planned ChatGPT platform redesign, and a deal letting customers apply Oracle cloud commitments to OpenAI products. Lower prices help defend the user base ahead of a public listing.

The risk is obvious — racing prices down while compute costs stay high pressures the path to profitability, the exact tension Nadella flagged. Watch whether OpenAI actually cuts, by how much, and whether it triggers a broader price war that benefits routing startups like OpenRouter and Vercel, which are booming as developers flock to the cheapest adequate model.