GPT-6 Luna (Low): benchmark results
Provider: OpenAI. Released 2026-09-22. Access: API.
Aggregate Rating 1663 ± 11, rank #479 of 2178 rated models, from 135 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MLCR-AA - Accuracy | 75 | Accuracy (%) | 82.3 |
| AA Long Context Reasoning | 74 | Accuracy (%) | 69.1 |
| CMDB-1500 (Image) | 89.33 | Accuracy (%), fixed denominator of 300 questions (invalid an | 66.7 |
| EnigmaForge - Intuition | 22.9 | Task success (%) on implicit-condition instances, where the | 66.7 |
| AA Humanity's Last Exam | 20.25 | Accuracy (%) | 66.1 |
| MLCR-AA | 18.89 | Overall Score (%) | 66 |
| AA MMMU-Pro | 75.95 | Accuracy (%) | 65.4 |
| AA CritPt | 2.57 | Accuracy (%) | 63.4 |
| AA GDP.pdf - Construction - Criterion Pass Rate | 65.25 | Criterion Pass Rate (%) | 62.4 |
| LisanBench | 0.04 | Mean Path Length / Current Maximum | 60 |
| DataBench | 30.3 | Score (%) | 56.3 |
| CMDB-1500 (Text) | 76.58 | Accuracy (%), fixed denominator of 1,200 questions (invalid | 55.6 |
Interactive version: theaggregate.ai/model?slug=gpt-6-luna-low · How It Works · Data refreshed daily, snapshot 2026-11-07.