OpenAI Cuts GPT-5.6 Luna Price 80% as It Gives 100,000 Researchers Free API Access
GPT-5.6 Luna drops to $0.20/$1.20 per million tokens and Terra falls 20%, while OpenAI opens free frontier-model access to scientists and engineers through 2027. The move lands the same day DeepSeek ships V4.
OpenAI cut API pricing on its two cheaper GPT-5.6 tiers on July 30. GPT-5.6 Luna, the budget model, drops 80% — from $1/$6 per million input/output tokens to $0.20/$1.20. GPT-5.6 Terra, the mid-tier, falls 20%, from $2.50/$15 to $2/$12. GPT-5.6 Sol, the flagship, stays at $5 input and $30 output per million tokens — unchanged.
Alongside the price cut, OpenAI is giving roughly 100,000 researchers, scientists, and engineers free access to its frontier models through 2027, explicitly targeting academics who’ve historically been priced out of running large-scale experiments on the newest models.
For anyone already on a ChatGPT Plus or Codex subscription, the practical effect is more usage at the same monthly price: Terra and Luna now consume fewer credits per request within existing subscription tiers, even though the subscription price itself hasn’t moved. OpenAI also upgraded Auto-review in the ChatGPT app and in Codex CLI to run on GPT-5.6 Luna, betting the cheaper model is now good enough for review-tier tasks that used to require a pricier one.
The timing is the real story. DeepSeek shipped V4 the same day, claiming improved coding and debugging performance, and OpenAI’s cut directly undercuts what Chinese labs have been using as their main lever — price. Luna at $0.20 per million input tokens is now cheap enough to compete on cost with open-weight alternatives that customers can self-host, not just other hosted APIs. That’s a different competitive fight than the one OpenAI was in a year ago, when differentiation was purely about capability.
It also signals where OpenAI thinks its margin can absorb pressure. Cutting Luna 80% while leaving Sol untouched preserves pricing power at the top of the stack — where enterprise customers pay for frontier capability regardless of cost sensitivity — while conceding the commodity, high-volume tier to whoever can match the price. Expect Anthropic and Google to face pressure to respond in kind on their own budget-tier models (Claude Haiku, Gemini Flash) if Luna’s cut pulls meaningful volume away from them.
For developers building on the API: if your workload was routing to Terra for cost reasons, it’s worth re-testing against Luna at the new pricing — an 80% cut changes the calculus on what’s worth running on the cheapest tier versus what still needs the mid-tier’s extra headroom.