When deploying GPT-5.6-Luna with short context in the Global environment, the pricing shown in the deployment flow lists output tokens at $6 per million. However, the published announcement lists a lower price of $1.20 per million output tokens. Has anyone confirmed which price is currently correct, and when the lower rate will take effect?
3 Answers
Others appear to be seeing the same discrepancy: the announcement describes lower pricing, while actual deployment pricing remains higher. It may be a regional rollout or catalog update that has not propagated yet.
I checked the pricing API for a Global short-context Luna deployment, and it still showed $1.00 per million input tokens, $0.10 for cached input, $1.25 for cache writes, and $6.00 per million output tokens. That suggests the lower advertised output price has not rolled out to this deployment yet.
The pricing displayed on the general website may not be reliable for a specific region or deployment type. The retail pricing API is a better source for the rate actually being applied to your account and location.
That matches what I’m seeing too. The published announcement makes it sound like the lower rate should already be active, but billing and the API still show the higher output price.