OpenAI GPT-6 Sol & Luna Launch - Factual Errors on User-Flagged Conversations: leaderboard
Metric: Answers with any factual error (%). Source: openai.com. Saturation forecast: Estimated already saturated. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 (High) | 3.9 |
| 2 | GPT-6 (Max) | 3.91 |
| 3 | GPT-6 (xHigh) | 3.99 |
| 4 | GPT-6 (Medium) | 4.41 |
| 5 | GPT-6 Sol (xHigh) | 4.52 |
| 6 | GPT-6 Sol (Max) | 4.57 |
| 7 | GPT-6 Sol (High) | 5.14 |
| 8 | GPT-6 (Low) | 6.26 |
| 9 | GPT-6 Sol (Medium) | 6.86 |
| 10 | GPT-6 Luna (Max) | 7.56 |
| 11 | GPT-5.6 Sol (xHigh) | 8.44 |
| 12 | GPT-5.6 Sol (Max) | 8.5 |
| 13 | GPT-6 Luna (xHigh) | 10.15 |
| 14 | GPT-5.6 Sol (High) | 10.75 |
| 15 | GPT-6 Sol (Low) | 11.42 |
Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-6-sol-luna-launch-factual-errors-on-user-flagged-conversations · How It Works · Data refreshed daily, snapshot 2026-09-24.