OpenAI Launches GPT-5.6 After Trump Administration Lifts Restrictions

The full GPT-5.6 family — Sol, Terra, and Luna — goes public after weeks of government-mandated limited preview, setting new benchmarks in coding and agentic reasoning.

GPT-5.6 is now generally available. OpenAI launched its full model family — flagship Sol, balanced Terra, and cost-efficient Luna — on Thursday after the Trump administration lifted restrictions that had kept the models in limited preview since late June.

Sol sets new state-of-the-art results across coding, cybersecurity, and science. On the Artificial Analysis Coding Agent Index it scores 80 — 2.8 points above Anthropic's Fable 5 — while using less than half the output tokens and costing about one-third less. On Agents' Last Exam, which measures long-running professional workflows across 55 fields, Sol scored 53.6, beating Fable 5 by 13.1 points.

The model family introduces "ultra" mode, which coordinates four AI agents in parallel by default to handle complex tasks faster than a single model instance. GPT-5.6 also scored 92.2% on BrowseComp, a benchmark for agentic browsing and research, and 62.6% on OSWorld 2.0 for computer-use tasks.

The launch follows weeks of regulatory tension. OpenAI previewed GPT-5.6 on June 25 to government-approved entities only, a staggered rollout the company said was not its preferred approach. The Trump administration signed off on the full release this week after additional testing by the Commerce Department's Center for AI Standards and Innovation. The White House disputes the "green light" framing, noting that Trump's June 2 AI executive order explicitly bars mandatory federal licensing for AI model releases.

The launch is the top story on Hacker News, where the discussion has drawn over 850 points and 630 comments.

Sources: The Verge, Hacker News