GPT 5.6 Public Release

Thirteen days ago, GPT-5.6 was available to the U.S. government and almost no one else.

Today it's available to everyone.

OpenAI released Sol, Terra, and Luna to the public this morning after the Commerce Department's Center for AI Standards and Innovation completed an additional round of testing. That clearance was the gate. Not the training run, not the eval suite, not the infrastructure. A federal review.

I think we're watching the release cycle change shape in real time.

For most of the last decade, a frontier model shipped when the lab decided it was ready. The constraint was capability. Now there's a second constraint sitting upstream of launch day, and it isn't optional, and it doesn't move on your roadmap's schedule.

The pricing tells you what the labs do with that constraint once they clear it. Sol runs $5 per million input tokens and $30 output. Terra is $2.50 and $15. Luna lands at $1 and $6 — the cheapest tier OpenAI has ever shipped. Terra reportedly matches GPT-5.5 at roughly half the cost.

So capability keeps compounding and the price floor keeps dropping, but the calendar now belongs partly to a regulator.

If you're planning enterprise AI deployments, that changes what you're actually managing. Model risk was a technical question. It's becoming a scheduling question. The vendor timeline you built your Q4 around now has a dependency you don't control and can't escalate.

Worth asking your vendors directly: what's your regulatory clearance path, and what happens to my rollout if it slips?

What's your read — is pre-clearance a temporary artifact of the export-control moment, or the permanent shape of frontier releases?