Benchmarks · coding
DuelLab GameBench 2
Models write game-playing programs from a rules description; the programs are compiled and play head-to-head tournaments on hidden abstract strategy games. 0-100 score, penalised for programs that fail to run.
71 models · 173 results · updated 20261026 · benchmarks.duellab.org ↗
Leaderboard
Opus 5.5 xhigh
More from this benchmark
5 published measuresScore vs any published measure5 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | |||
|---|---|---|---|
| Opus 5.5 xhigh, fallback | 3,846 | 163 | 79 |
| GPT-6 Astra xhigh | 3,804 | 188 | 96 |
| Fable 5.1 xhigh, fallback | 5,197 | 288 | 125 |
| GPT-6.1 Sol xhigh | 3,724 | 259 | 105 |
| Sonnet 5.5 xhigh, fallback | 3,585 | 190 | 313 |
| Gemini 4 Argon high | 3,579 | 341 | 168 |
| Opus 5 xhigh | 5,430 | 835 | 377 |
| Kimi K3.1 max | 3,294 | 456 | 338 |
| GPT-5.6 Sol xhigh | 3,107 | 681 | 300 |
| GPT-6 Sol xhigh | 3,150 | 698 | 240 |
| Muse Wonder (Max) xhigh | 2,886 | 667 | 535 |
| Fable 5 xhigh | 2,908 | 873 | 307 |
| GLM 5.3 max | 2,780 | 412 | 380 |
| Grok 4.7 medium | 2,860 | 512 | 214 |
| MiMo V2.6 Pro high | 2,741 | 961 | 386 |
| DeepSeek V4.1 Flash 1017 max | 2,577 | 942 | 567 |
| Grok 4.6 medium | 2,460 | 841 | 787 |
| Muse Spark 1.3 xhigh | 2,554 | 1,005 | 529 |
| Qwen 3.8 Max xhigh | 2,385 | 887 | 816 |
| DeepSeek V4.1 Flash max | 2,360 | 971 | 757 |
| Qwen 3.8 2.4T-A95B xhigh | 2,325 | 923 | 840 |
| Gemini 3.1 Pro high | 2,375 | 1,162 | 551 |
| Gemini 3.8 Flash high | 2,387 | 1,241 | 460 |
| Gemini 3.6 Flash high | 4,359 | 2,326 | 1,492 |
| Hy4 high | 2,053 | 1,062 | 973 |
All results
173 rows| # | Model | Effort | DuelLab GameBench 2 |
|---|---|---|---|
| 1 | Opus 5.5 | xhigh | 95.1 ±7.6 |
| 2 | GPT-6 Astra | xhigh | 94.4 ±6.0 |
| 3 | Fable 5.1 | xhigh | 94.4 ±5.1 |
| 4 | Opus 5.5 | medium | 93.0 ±4.3 |
| 5 | GPT-6.1 Sol | xhigh | 92.7 ±11.1 |
| 6 | Sonnet 5.5 | xhigh | 91.8 ±9.3 |
| 7 | Gemini 4 Argon | high | 90.1 ±13.1 |
| 8 | GPT-6 Astra | medium | 89.6 ±9.9 |
| 9 | GPT-6.1 Sol | medium | 87.5 ±10.6 |
| 10 | Fable 5.1 | low | 86.6 ±9.5 |
| 11 | Opus 5 | xhigh | 86.1 ±12.9 |
| 12 | Kimi K3.1 | max | 85.7 ±9.3 |
| 13 | GPT-6 Astra | low | 85.1 ±11.0 |
| 14 | Fable 5.1 | medium | 84.8 ±14.6 |
| 15 | GPT-5.6 Sol | xhigh | 81.1 ±6.8 |
| 16 | Gemini 4 Argon | low | 81.0 ±16.5 |
| 17 | Kimi K3.1 | low | 80.8 ±26.0 |
| 18 | GPT-6 Sol | xhigh | 80.6 ±42.0 |
| 19 | Muse Wonder (Max) | xhigh | 78.8 ±15.5 |
| 20 | Gemini 4 Argon | medium | 78.7 ±14.6 |
| 21 | Opus 5.5 | low | 78.2 ±19.6 |
| 22 | GPT-6.1 Sol | low | 77.8 ±21.5 |
| 23 | Fable 5 | xhigh | 76.9 ±15.4 |
| 24 | Opus 5 | medium | 76.5 ±22.3 |
| 25 | GPT-6 Sol | medium | 76.4 ±27.1 |
| 26 | GLM 5.3 | max | 73.5 ±46.5 |
| 27 | Grok 4.7 | medium | 73.4 ±46.5 |
| 28 | Grok 4.7 | xhigh | 73.2 ±46.7 |
| 29 | MiMo V2.6 Pro | high | 73.2 ±22.2 |
| 30 | GLM 5.3 | high | 72.0 ±20.4 |