OpenAI Considers Drastic Price Cuts as AI Token War With Anthropic Heats Up

OpenAI is preparing to slash the prices it charges for AI tokens, according to a ¹ exclusive citing people familiar with the matter. The move is explicitly aimed at winning customers from rival Anthropic, whose coding tool Claude Code has been devouring market share among software engineers.

The company is weighing significant cuts to what it charges for tokens, the unit of measurement artificial‑intelligence firms use to bill for their products, the WSJ reported, noting the discussions are still in flux. OpenAI is acting preemptively -- anticipating similar cuts from Anthropic in what increasingly looks like an unavoidable price war.

OpenAI CEO Sam Altman acknowledged the pressure directly at a recent event. I think we'll have a lot of ways we can help people get more value for less spend, Altman said, according to Quartz. He called costs a huge issue for business customers.

The urgency behind the price cut discussions stems from a dramatic shift in competitive dynamics. Anthropic's Claude Code -- a coding tool that lets developers ship complex changes through natural language -- went viral among software engineers, driving what the ¹ described as mind‑blowing growth that propelled the five‑year‑old startup into its first profitable quarter.

That growth carried Anthropic past a significant milestone: in May, the company closed a $65 billion Series H round at a $965 billion valuation, according to Quartz, edging past OpenAI's $852 billion valuation for the first time. OpenAI has since made its own coding tool, Codex, a central focus of the company -- acquiring a cloud startup to strengthen it.

The subscriber battlefield tells its own story. OpenAI charges consumers in tiered subscriptions of $8, $20, and $100‑plus monthly for GPT‑5.5 access, while CNBC reports Anthropic charges $17 monthly (with annual commitment) for Claude Pro and $100‑plus for Claude Max.

Anthropic launched Claude Fable 5 -- its first publicly available Mythos‑class model -- on June 9 with pricing that significantly raised the ceiling. Barron's reports Fable 5 costs $10 per 1 million input tokens and $50 per 1 million output tokens -- exactly double the $5/$30 pricing of GPT‑5.5.

The price war arrives just as corporations are confronting a phenomenon IBM Consulting SVP Neil Dhar calls tokenmaxxing -- the organizational push to use as much AI as possible, as fast as possible, driven by two years of competitive anxiety, where the pressure to adopt AI outpaced serious consideration of long‑term return on investment, Barron's reported.

The consequences are already visible. Uber recently capped employee usage of AI tools like Claude Code to control costs, and multiple technology companies have faced unexpectedly high AI bills, according to Bloomberg. The spending spree has some companies already maxed out on their AI budgets.

Drastic price cuts could potentially erode the profit margins of both companies, which already lose billions of dollars because of the enormous cost for computing resources, the WSJ noted -- underscoring the high‑stakes gamble of slashing prices while bleeding cash on compute.

Both companies are racing toward initial public offerings that will test whether public markets are willing to value money‑losing AI labs at trillion‑dollar‑plus levels. OpenAI confidentially filed for its IPO on June 8, while CNBC reports Anthropic had already filed. A price war adds an awkward dimension to both S‑1 narratives.

If they sacrifice too much margin to maintain growth, investors might be concerned when they reach the public market, ³ warned. The outlet also noted that the WSJ reported on April 28 that OpenAI is missing its own revenue targets and is worried about meeting future spending contracts -- a report OpenAI disputed.

⁴ columnist Chris Bryant put it bluntly: None of this bodes well for both companies' forthcoming initial public offerings. OpenAI has committed to roughly $1.4 trillion in spending contracts, according to Barron's -- creating immense pressure to maintain revenue growth even as prices fall.

For developers building on these APIs, a price war means near‑term cost relief -- but it also signals a period of pricing instability. If both companies engage in aggressive discounting, per‑token costs could drop substantially within weeks. The key question is whether cuts apply to premium tiers (GPT‑5.5, Claude Fable 5) or trickle down to cheaper models that most builders actually use in production.

The structural risk is that sustained below‑cost pricing forces one competitor to cut corners -- either on model quality, safety guardrails, or infrastructure investment. Builders with multi‑model architectures (routing between OpenAI and Anthropic based on task) are best positioned to weather the volatility. Those locked into a single provider should watch for contract renegotiation windows and evaluate backup options.

The competitive dynamics also suggest that coding tools -- Codex and Claude Code -- will be the primary battleground. Both companies see developer tooling as their growth engine, and pricing in this segment may move faster than consumer or enterprise tiers.