FORTE - Pass^3: leaderboard
Metric: Pass^3 (%, task solved in all three runs). Source: agi-eval-official.github.io. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.8 | 69.95 |
| 2 | GPT-5.5 | 68.85 |
| 3 | Claude Opus 4.7 | 68.85 |
| 4 | GLM-5.2 | 63.93 |
| 5 | Claude Opus 4.6 | 63.39 |
| 6 | GLM-5.1 | 61.2 |
| 7 | Gemini 3.1 Pro (Preview) | 60.11 |
| 8 | LongCat 2.0 | 59.02 |
| 9 | Claude Sonnet 4.6 | 56.83 |
| 10 | GPT-5.4 | 56.83 |
| 11 | MiniMax-M3 | 56.28 |
| 12 | DeepSeek V4 Flash | 56.28 |
| 13 | DeepSeek V4 Pro | 54.64 |
| 14 | Kimi K2.6 | 46.45 |
| 15 | Qwen 3.6 Plus | 45.9 |
Interactive version: theaggregate.ai/benchmark?slug=forte-pass-pow-3 · How It Works · Data refreshed daily, snapshot 2026-09-19.