OpenAI slashes GPT-5.6 Luna price 80% in AI pricing war

OpenAI cut the price of its cheapest GPT-5.6 model by 80% three weeks after launch, dropping Luna to 20 cents per million input tokens — a move that pressures margins across the AI industry while expanding the addressable market for high-volume automation.

"Under-the-hood improvements should be passed back to our users and customers in the form of more widely available, cost-efficient intelligence," five members of OpenAI's technical staff wrote in a July 29 engineering post detailing the optimizations that enabled the cuts.

The reductions put Luna at 20 cents per million input tokens and $1.20 per million output tokens, down from $1 and $6 at its July 9 launch. Terra, the mid-tier model, fell 20% to $2 and $12. Sol, the flagship, held at $5 and $30. The efficiency gains — a 20% reduction in end-to-end serving costs and a 15% improvement in token-generation efficiency — came from rewriting production GPU kernels and redesigning Sol's speculative-decoding draft model, OpenAI said.

The pricing shift arrives as enterprise customers scrutinize AI spending. Uber burned through its entire annual AI budget in four months, and Amazon's engineering organization capped AI spending after cost overruns. OpenAI shipped hard spend limits for API organizations on July 22, letting administrators set monthly ceilings. Both OpenAI and Anthropic have filed confidential IPO prospectuses with the SEC, adding pressure to demonstrate durable business models.

Where Luna and Terra sit against Anthropic

At 20 cents input and $1.20 output, Luna undercuts Anthropic's cheapest published model, Haiku 4.5, by a factor of five on input and roughly four on output. Terra's new rate of $2 and $12 sits below the $3 and $15 that Claude Sonnet 5 is scheduled to charge once its introductory rate of $2 and $10 lapses on Aug. 31. At the top of both lineups, Sol still costs more on output than Opus 5, which Anthropic prices at $5 and $25.

OpenAI also replaced Priority Processing with Fast mode, matching Anthropic's product name and terms: up to 2.5 times standard speed at twice the price. Fast-mode rates are $10 and $60 for Sol, $4 and $24 for Terra, and 40 cents and $2.40 for Luna. Both companies now charge a 10% uplift for region-pinned inference when customers require data residency.

What the cuts mean for AI buyers

The cheaper tiers are where high-volume production traffic lives: classification, extraction, request routing, first-pass drafting, and the long agent loops where one user instruction can trigger dozens of model calls. A five-fold cut on the tier absorbing that volume changes the arithmetic on which workloads are worth automating.

OpenAI caps tool output at 10,000 tokens by default and keeps model-visible history append-only, so agent loops resending instructions and tool definitions at every step hit the prompt cache instead of paying full input rates. On a task requiring 30 model requests, that is 30 chances to avoid recomputing the same prefix. For a team already routing bulk work to Luna, the same calls now cost a fifth of what they did, with a ceiling they can set in the dashboard.

The cuts also benefit ChatGPT Work and Codex subscribers. OpenAI said the lower prices are reflected in how usage is counted against paid subscriptions, meaning Terra and Luna users can complete more tasks before hitting limits. Subscription prices remain unchanged.

OpenAI's own research found staff using ChatGPT well beyond their job titles, suggesting the lower costs could widen internal adoption. With Sol's price unchanged, the live decision stays where OpenAI has put it since the family shipped: which tier each request requires. Cheaper Chinese open-weight models have sharpened the competitive environment, pushing OpenAI and Anthropic to make the case that their products are worth the premium.

This article is for informational purposes only and does not constitute investment advice.