Opus Magnum Campaign
Games · 2026-06-17
Campaign benchmark score across all 36 Opus Magnum puzzles, normalized to human-best sum scores.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Sol | 70.9 |
| Claude Fable 5 | 60.2 |
| GPT-5.5 | 53.9 |
| GPT-5.6 Terra | 47.7 |
| Gemini 3.1 Pro Preview | 31.1 |
| Sonnet 5 | 26.5 |
| GLM-5.2 | 25.4 |
| Gemini 3.5 Flash | 24.5 |
| Opus 4.8 | 22.8 |
| Grok 4.5 | 18.2 |
| GPT-5.6 Luna | 17.8 |
| Gemini 3 Flash Preview | 15.4 |
| DeepSeek-V4-Pro | 13.4 |
| Sonnet 4.6 | 12.4 |
| Kimi K2.7 Code | 10.9 |
| GLM-5.1 | 10.8 |
| GPT-5.4 Mini | 9.2 |
| GPT-5.4 Nano | 6.5 |
| Kimi K2.6 | 5.5 |
| Grok 4.3 | 0.0 |