OpenAI launched GPT-5.6 on Friday, less than a day after reports surfaced that the Trump administration had asked the company to delay the release.
The suite ships in three tiers: Sol, the flagship; Terra, pitched at "high-volume work"; and Luna, billed as "fast and affordable" for everyday use. OpenAI says the models perform well in coding, cybersecurity, and biology, and are designed to maintain focus during long-horizon agentic tasks. Sol is priced at $5 per million input tokens and $30 per million output tokens. What OpenAI has not explained is what, if anything, it changed in response to the administration's request, or whether the 24-hour gap was the accommodation itself.
The administration's ability to informally stagger a major model launch is a more consequential story than the specs. It means political actors can influence AI deployment timelines without a legal mandate, a formal regulatory process, or a public explanation. That precedent will outlast GPT-5.6.
The launch went ahead anyway, which suggests the ask had a short shelf life, or OpenAI decided it did.