OpenAI
GPT-6 Luna benchmark results
5 results on 5 benchmarks from 2 sources.
Also listed as: GPT-6 Luna (Max) · gpt-6-luna_max · GPT-6 Luna (max)
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1442 | 91 | Overall, style control on; 72504 votes; 95% CI 1438–1447“GPT-6 Luna (Max)” | Official leaderboard | Oct 22, 2026 |
| LMArena WebDev | 1578 | 38 | Overall; 24237 votes; 95% CI 1572–1584“GPT-6 Luna (Max)” | Official leaderboard | Oct 22, 2026 |
| LMArena Vision | 1240 | 66 | Overall, style control on; 3662 votes; 95% CI 1229–1251“GPT-6 Luna (Max)” | Official leaderboard | Oct 19, 2026 |
| GPQA Diamond | 90.5% | 39 | Epoch AI run, accuracy; effort: max | Independent run | Sep 22, 2026 |
| AIME (OTIS Mock AIME 2024–2025) | 98.9% | 20 | Epoch AI run, OTIS Mock AIME 2024–2025; effort: max | Independent run | Sep 22, 2026 |
Where GPT-6 Luna ranks
Primary setting, one best-sourced score per model (official / independent first); neighbours in brackets.
| Benchmark | Score | Rank | Nearby models | #1 |
|---|---|---|---|---|
| AIME (OTIS Mock AIME 2024–2025) | 98.9% | #20 / 200 | Grok 4.6 99.2% · Claude Haiku 5.5 98.9% · Gemini 3.8 Flash 98.9% · Claude Opus 5 98.9% | Gemini 4 Argon100.0% |
| GPQA Diamond | 90.5% | #35 / 217 | Claude Sonnet 5 90.5% · Claude Opus 4.6 90.5% · GLM-5.3-Flash 90.2% · Claude Opus 4.7 90.2% | Gemini 4 Argon96.4% |
| LMArena Text | 1442 | #82 / 392 | Qwen 3.6 Plus 1443 · ChatGPT-4o Latest (20250326) 1443 · Qwen3.5 397B-A17B 1442 · Grok 4.3 Beta 1442 | Gemini 4 Argon1517 |
| LMArena WebDev | 1578 | #34 / 135 | DeepSeek-V4-Pro (High, 20260813) 1582 · DeepSeek-V4-Flash 1582 · Step 5 Preview 1568 · Claude Opus 4.7 1558 | Claude Opus 5.51807 |
| LMArena Vision | 1240 | #58 / 154 | Grok 4.3 Beta 1242 · Gemma 4 26B A4B 1242 · ChatGPT-4o Latest (20250326) 1239 · Gemini-3.1-Flash-Lite-Preview 1235 | Claude Fable 51309 |
Other OpenAI models
GPT-5GPT-5.2o3GPT-4.1GPT-5.4o4-miniGPT-4 Turbo (Nov 2023)GPT-4o (May 2024)GPT-5 miniGPT-5.1GPT-4.1 miniGPT-4o mini
FAQ
Where does GPT-6 Luna rank best?
AIME (OTIS Mock AIME 2024–2025): #20 of 200 (98.9%); #1 there is Gemini 4 Argon (100.0%).
Where do these GPT-6 Luna scores come from?
From Epoch AI Benchmarking Hub and LMArena leaderboard dataset. Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.