OpenAI ships GPT-5.6, and the government wanted a look first

OpenAI has released GPT-5.6 in three tiers. Soul is the flagship. Terra matches the last model's capability at half the price. Luna is the cheap, fast option built for high volume work.

The headline is not the polish, it is the security. OpenAI is calling GPT-5.6 its strongest cybersecurity model yet, good enough at finding software vulnerabilities that the US government asked to audit it before any wide release. For now it goes to a vetted set of partners only.

Why it matters: a model that finds bugs faster than people helps defenders and attackers alike, which is exactly why the release is gated. It is the first time a frontier lab has slowed its own launch for a government security review.

Get the free daily Future Technology briefing at futuretechnologyhq.com/newsletter.