![\"ARC](\"https:\/\/arcprize.org\/media\/images\/badges\/arc-prize-verified-badge-dark-color.svg\"){kind=icon}
<\/p>\n<\/p>\n
OpenAI\u00b7Jul 9, 2026\u00b73 models\u00b715 reasoning variants<\/p>\n
GPT-5.6 Sol is the standout model of the GPT-5.6 family. Sol at max reasoning effort is the only performant model (as of July 2026) averaging 13.33% on Public and 7.78% on Semi-Private. It is the first model to win an ARC-AGI-3 public game (ft09, 87%). Sol is able to read an unfamiliar scene correctly and in the game’s own vocabulary. It treats a failed hypothesis as a reason to re-plan rather than thrash. Most agent failures are upstream of the code they write or the action they take. Sol is able to perform on ARC-AGI not because it executes better, but because it correctly orients itself in a new environment first.<\/p>\n
ARC-AGI 3 leaderboard<\/p>\n
GPT-5.6 SolGPT-5.6 TerraGPT-5.6 Luna<\/p>\n
ARC-AGI-1ARC-AGI-2ARC-AGI-3<\/p>\n
Verified scoresModelVariantARC-AGI-1ARC-AGI-2ARC-AGI-3Sol<\/a>Max<\/p>\n 96.5%<\/p>\n 92.5%<\/p>\n 7.8%<\/p>\n Extra High<\/p>\n 97.5%<\/p>\n 90.0%<\/p>\n 7.0%<\/p>\n High<\/p>\n 97.0%<\/p>\n 85.4%<\/p>\n 2.1%<\/p>\n Medium<\/p>\n 92.5%<\/p>\n 67.1%<\/p>\n 1.1%<\/p>\n Low<\/p>\n 74.5%<\/p>\n 42.5%<\/p>\n 0.3%<\/p>\n Terra<\/a>Max<\/p>\n 96.5%<\/p>\n 83.9%<\/p>\n 0.8%<\/p>\n Extra High<\/p>\n 94.0%<\/p>\n 74.2%<\/p>\n 0.7%<\/p>\n High<\/p>\n 92.0%<\/p>\n 67.1%<\/p>\n 0.5%<\/p>\n Medium<\/p>\n 77.0%<\/p>\n 37.5%<\/p>\n 0.1%<\/p>\n Low<\/p>\n 60.2%<\/p>\n 18.8%<\/p>\n 0.0%<\/p>\n