OpenAI cut GPT-5.6 Luna pricing by 80% and GPT-5.6 Terra by 20%, effective July 30. Luna now costs $0.20 per million input tokens and $1.20 per million output tokens. Terra dropped to $2.00 input and $12.00 output. Sol pricing remains unchanged. The repricing came three weeks after the GPT-5.6 family launched on July 9.
OpenAI attributed the cuts to inference efficiency improvements, including kernel optimizations that reduced serving costs by 20% and token-generation efficiency gains exceeding 15%. The company said GPT-5.6 Sol, its most capable model, autonomously rewrote production kernels and ran hundreds of optimization experiments within a human-led process.
The Speed of Repricing
Three weeks from launch to an 80% price cut is unprecedented in OpenAI’s history. Previous model generations typically held launch pricing for months before reductions. GPT-4o launched at $5.00/$15.00 per million tokens and still lists at $2.50/$10.00 after over two years. GPT-5.6 Luna’s trajectory is fundamentally different.
Axios reported the cuts reflect competitive pressure from Chinese open-weight models, including Moonshot AI’s Kimi K3, which are closing capability gaps while offering cheaper alternatives. OpenAI framed the reduction as efficiency-driven rather than competitive.
Where Luna Sits Now
At $0.20 per million input tokens, Luna is now cheaper than GPT-4o-mini ($0.15 input but $0.60 output) on output-heavy workloads and approaches GPT-5-nano ($0.05/$0.40) territory while delivering substantially higher capability. OpenAI claims Luna outperforms Anthropic’s Fable 5 on Agents’ Last Exam at an estimated cost per task “nearly 99% lower,” according to the GPT-5.6 announcement.
Michele Catasta, President and Head of AI at Replit, told OpenAI: “GPT-5.6 Luna is the closest we’ve come to intelligence too cheap to meter. I’ve never seen a model this affordable be this powerful.”
OpenAI also introduced Fast mode for GPT-5.6 Sol in the API, replacing Priority Processing. Fast mode delivers up to 2.5x faster speeds at twice the standard price with no change in intelligence. Existing API requests tagged “priority” automatically route to Fast mode.
The Pricing Power Question
The cuts arrive days after OpenAI CFO Sarah Friar disclosed that July annualized recurring revenue exceeded the entire Q2 total. Revenue growth and aggressive repricing happening simultaneously suggests OpenAI believes volume gains from lower prices will outweigh margin compression, or that maintaining market share against cheaper open-weight alternatives requires sacrificing pricing power now. The answer determines whether frontier model companies can sustain the infrastructure spending their business models require.