OpenAI vs Anthropic: API token price cuts and the race

OpenAI is weighing cuts to per-token API prices to win customers from Anthropic. The Wall Street Journal published details of internal discussions that suggest price is becoming a front-line battleground between the two companies.

OpenAI has long set the tone for commercial model pricing, charging per-token fees that scale with model capability and context window. The company is now exploring lower per-token rates to blunt Anthropic's push for enterprise contracts and volume-based deals, according to the published account. Executives inside OpenAI are said to be debating how steep a cut would be sustainable given compute costs and margins.

What OpenAI is considering

Sources described options ranging from modest discounts on current models to deeper cuts tied to volume commitments or multi-year contracts. Any change would affect billing for both short-form requests and long-context use cases where token consumption is high. Lower per-token rates would make extended-context calls and token-heavy applications, such as long-document analysis and chat transcripts, materially cheaper for customers.

Reducing token prices is not the only lever. OpenAI could adjust batch pricing, offer credits or rebates for large customers, or bundle usage with enterprise support and fine-tuning credits. Each approach sacrifices some near-term revenue but can increase stickiness and usage across developer ecosystems. Internal deliberations reportedly weigh those trade-offs against the companys hardware and cloud spending, which remain a dominant cost for high-capacity models.

Publicly, OpenAI has not announced a change. The company continues to position newer models at premium price points, while offering older versions at lower rates. Any formal price move would likely be coordinated with partners and accompanied by messaging about improved efficiency or performance per dollar.

Anthropic and the competitive backdrop

Anthropic has positioned its Claude models as a safety- and enterprise-focused alternative, and its commercial pricing strategy has leaned into predictable, enterprise-friendly contracts. For some customers, Anthropics pricing and service terms have been attractive enough to encourage migration from other providers.

Competition between model providers has intensified as enterprises press for predictable costs and higher throughput. Cloud vendors and model operators are experimenting with discounts, committed-use deals and new billing units to win larger customers. For buyers, lower per-token costs reduce friction for scaling production systems and for running token-heavy workloads.

Price is only one axis of competition. Model performance, integrations, service SLAs and regulatory compliance also influence procurement decisions. Firms choosing among providers weigh these factors against unit economics for large-scale deployments.

Why it matters

A move by OpenAI to cut per-token API prices would shift the economics of running large-scale, token-intensive applications and could pressure competitors to match discounts or redesign pricing tiers. Enterprise customers and high-volume developers stand to benefit most from lower token costs, while providers must reconcile price cuts with rising compute expenses and margin pressures.