OpenAI Cuts GPT-5.6 Luna Price 80%, Terra 20%, as Enterprise Cost Pressure Mounts

OpenAI is cutting prices on two of its GPT-5.6 models roughly three weeks after their public release, as enterprise customers grow increasingly sensitive to AI compute costs.

The company announced Thursday it is reducing the price of GPT-5.6 Terra, its mid-tier model, by 20% to $2 per million input tokens and $12 per million output tokens. GPT-5.6 Luna, the fastest and cheapest model in the lineup, gets an 80% price cut, dropping to $0.20 per million input tokens and $1.20 per million output tokens. Pricing for GPT-5.6 Sol, the most capable model in the series, remains unchanged.

The GPT-5.6 family launched with three tiers: Sol at the top, Terra in the middle, and Luna as the speed-optimized option. OpenAI framed the cuts as part of a continuing efficiency push. "Our strategy remains focused on advancing both capability and efficiency so each generation of intelligence can accomplish more work at a lower cost," the company said in a release.

Why now

Enterprises that once deployed AI models liberally — an approach some in the industry nicknamed "tokenmaxxing" — have grown more disciplined about spending as bills scale into the billions of dollars. That shift has coincided with rapid gains from Chinese open-weight models, narrowing the performance gap with proprietary U.S. systems.

Moonshot AI's Kimi K3, an open-weight model released earlier this month, reportedly outperformed leading American models on several industry benchmarks, according to the company. The release prompted a quick response across Silicon Valley. Anthropic followed with Claude Opus 5, which it says matches Claude Fable 5 on coding and knowledge-work benchmarks at half the price. Google debuted three new models this month, including Gemini 3.6 Flash, which the company claims is cheaper per task than Kimi K3 and other Chinese models. Microsoft CEO Satya Nadella repeatedly emphasized cost-effective models during the company's earnings call this week, days after Microsoft released what it described as a cheap-but-performant cybersecurity model.

None of the competitive benchmark claims from Moonshot, Anthropic, or Google have been independently verified by third-party evaluators as of this writing.

What this means

The price cuts signal that the AI industry's competitive battleground has shifted from raw capability to cost-per-task economics. OpenAI still charges a premium for Sol, preserving margin on its flagship model while using Terra and Luna as volume plays aimed at enterprise workloads where output-token costs compound quickly at scale — a chatbot or coding agent running millions of daily requests feels an 80% cut far more than a research lab running occasional frontier queries.

The timing, three weeks after launch, is notable: it suggests OpenAI is responding to real-time competitive pressure rather than following a pre-planned pricing roadmap. With Anthropic, Google, and Moonshot AI all pushing cheaper alternatives simultaneously, buyers now have genuine leverage to negotiate or switch providers based on price alone — something that wasn't true a year ago when frontier capability, not cost, was the primary purchasing criterion. Expect further cuts across the industry as open-weight Chinese models keep pressuring margins on mid-tier and budget-tier offerings.