GPT-5.6: Frontier intelligence that scales with your ambition | OpenAI

OpenAI is releasing the GPT-5.6 family for general availability with three models: Sol, Terra, and Luna.
GPT-5.6 Sol is the flagship model and is positioned as the strongest overall option.
Sol achieves state-of-the-art results in coding, knowledge work, cybersecurity, and science while using fewer tokens and lower estimated cost.
A new “ultra” setting coordinates multiple agents in parallel to speed up the hardest tasks.
GPT-5.6 was trained to produce more useful work from each token, improving efficiency on long-horizon workflows.
On Agents’ Last Exam, Sol reaches a new high of 53.6 and beats Claude Fable 5 by 13.1 points in adaptive reasoning.
Terra and Luna are highlighted as lower-cost models that still outperform Fable 5 at about one-sixteenth the cost.
On the Artificial Analysis Intelligence Index, Sol comes within one point of Fable 5 while taking 61% less time and about half the estimated cost.
The models include stronger safeguards and more extensive prelaunch testing, including human red teaming, automated testing, expert partnerships, and layered protections.
GPT-5.6 Sol is described as the best coding model yet, with benchmark highs on Artificial Analysis Coding Agent Index, Terminal-Bench 2.1, and DeepSWE.

Your notes

Save this item to your library to add private notes.