OpenAI cut GPT-5.6 Luna API pricing up to 80% just three weeks after launch, alongside

OpenAI just slashed GPT-5.6 Luna API pricing by up to 80%, three weeks after the model launched. They also rolled out a “Fast mode” and improved token efficiency by more than 15%. That’s not a scheduled price drop. That’s a reaction.

When a company cuts prices that hard, that fast, it usually means customers pushed back. Enterprise buyers have been complaining about AI costs for months, especially as usage scales. If you’re running thousands of API calls a day, even small per-token charges add up fast. OpenAI clearly heard it. The original report frames this as a response to real market pressure, not a planned optimization.

The token efficiency gain matters too. If the model uses 15% fewer tokens to do the same job, your bill drops again, even before the price cut. Combined, you’re looking at meaningful savings if you’re already running GPT-5.6 Luna in production. If you’re not, this makes the cost case a lot easier to justify.

What it means for you

If you’re evaluating AI tools or already using OpenAI’s API, rerun your cost models. The math just changed. If you shelved a use case because the API cost didn’t pencil out, revisit it. If you’re using a competitor, check whether they match or beat this new pricing.

This is also a reminder that AI infrastructure costs are still moving fast. Locking into a single vendor or assuming stable pricing is risky. The kind of work we build into an AI command centre includes monitoring cost per task across models and switching when it makes sense, so you’re not stuck manually tracking every price change.

Bottom line: OpenAI blinked on price. That’s good for buyers, but it also signals the market’s still figuring out what sustainable AI economics look like.