GPT-6 Luna

OpenAIReleased 22 Sep 2026ClosedFlagship

Compare Announcement

166155–178Index score#27 of 229 models

4/ 8Core tests taken4 of 6 areas · plus 2 anchor tests

−33Versus the frontier at releasebest then: Claude Opus 5.5

$0.10/ $0.50Price per million tokensinput / output, list price

1.05MContext windowtokens it reads at once

Where it sitson the frontier

Claude Opus 4.6 reached this score 8 months earlier.

Test resultsby area · tick = best by any model

Reasoning & knowledgeevidence

GPQA Diamond

90%

Humanity's Last Exam

no result

Mathsevidence

FrontierMath T1-3

79%

Codingno evidence yet

SWE-bench Verified

no result

Terminal-Bench

no result

Agentsevidence

APEX-Agents

44%

Novel problemsevidence

ARC-AGI-2

59%

Long tasksno evidence yet

METR time horizon

no result

Independent results from the Epoch AI benchmarking hub, best reasoning setting. The core tests are shown here; anchor tests also inform the score and can be shown above.

What others sayrank on each leaderboard

The Model Indexindependent tests, our method#27of 219 · 166

Artificial Analysis Intelligence Indexoverall capability across ten evaluations#21of 207 · 38 LMArena · Codingcoding questions in chat#52of 383 · 1511 LMArena · Mathsmaths questions in chat#71of 373 · 1448 LMArena · Textoverall chat quality, judged by people#78of 388 · 1444 LMArena · Creative writingcreative writing#82of 386 · 1409 LMArena · WebDevbuilding web apps and front ends#28of 123 · 1582 LiveBench · Overalloverall, with fresh questions to limit contamination#44of 58 · 72

Further left is better. Leaderboards measure different things: LMArena is people voting blind between two answers, the others are test scores. Click any row for the source.