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.
77 models · 189 results · updated 20261103 · benchmarks.duellab.org ↗
Leaderboard
Opus 5.5 xhigh
More from this benchmark
5 published measuresScore vs any published measure5 measures: sub-scores, splits, counts
60 more models under All models in settings.
Breakdown
darker is better within each column · click a column to sort| Model | |||
|---|---|---|---|
| Opus 5.5 xhigh | 3,939 | 165 | 80 |
| Fable 5.1 max | 3,902 | 196 | 82 |
| GPT-6 Astra xhigh | 3,890 | 186 | 108 |
| GPT-6.1 Sol xhigh | 3,806 | 279 | 99 |
| Sonnet 5.5 xhigh | 3,667 | 202 | 313 |
| Gemini 4 Argon high | 3,628 | 371 | 181 |
| GPT-6.1 Terra xhigh | 3,408 | 429 | 343 |
| GPT-6.1 Terra Pro xhigh | 3,445 | 536 | 193 |
| Opus 5 xhigh | 5,497 | 902 | 401 |
| Kimi K3.1 max | 3,331 | 507 | 356 |
| GPT-5.6 Sol xhigh | 3,154 | 715 | 309 |
| GPT-6 Sol xhigh | 3,191 | 739 | 254 |
| Muse Wonder (Max) xhigh | 2,934 | 711 | 545 |
| Fable 5 xhigh | 2,944 | 927 | 311 |
| GLM 5.3 max | 2,796 | 453 | 401 |
| Grok 4.7 medium | 2,891 | 553 | 222 |
| MiMo V2.6 Pro max | 2,764 | 1,018 | 394 |
| DeepSeek V4.1 Flash 1017 max | 2,594 | 1,007 | 585 |
| Muse Spark 1.3 xhigh | 2,562 | 1,072 | 544 |
| Grok 4.6 medium | 2,444 | 941 | 803 |
| Qwen 3.8 Max xhigh | 2,401 | 959 | 824 |
| GLM 5.3 FlashX max | 2,527 | 1,138 | 519 |
| DeepSeek V4.1 Flash max | 2,448 | 1,062 | 678 |
| Qwen 3.8 2.4T-A95B xhigh | 2,349 | 996 | 837 |
| Gemini 3.1 Pro high | 2,412 | 1,219 | 551 |
All results
189 rows| # | Model | Effort | DuelLab GameBench 2 |
|---|---|---|---|
| 1 | Opus 5.5 | xhigh | 94.9 ±8.4 |
| 2 | Fable 5.1 | max | 94.1 ±4.2 |
| 3 | GPT-6 Astra | xhigh | 94.0 ±7.0 |
| 4 | GPT-6.1 Sol | xhigh | 91.9 ±11.7 |
| 5 | Opus 5.5 | medium | 91.6 ±5.6 |
| 6 | Sonnet 5.5 | xhigh | 91.3 ±10.2 |
| 7 | Gemini 4 Argon | high | 89.0 ±13.1 |
| 8 | GPT-6 Astra | medium | 88.6 ±11.0 |
| 9 | GPT-6.1 Sol | medium | 86.3 ±11.6 |
| 10 | GPT-6.1 Terra | xhigh | 85.7 ±11.8 |
| 11 | GPT-6.1 Terra Pro | xhigh | 84.8 ±27.9 |
| 12 | Fable 5.1 | low | 84.8 ±9.7 |
| 13 | Opus 5 | xhigh | 84.7 ±13.4 |
| 14 | Kimi K3.1 | max | 84.0 ±9.8 |
| 15 | GPT-6 Astra | low | 83.5 ±10.9 |
| 16 | GPT-6.1 Terra Pro | none | 82.9 ±14.8 |
| 17 | Fable 5.1 | medium | 82.9 ±14.6 |
| 18 | GPT-5.6 Sol | xhigh | 79.6 ±6.6 |
| 19 | GPT-6.1 Terra | none | 79.4 ±15.3 |
| 20 | GPT-6 Sol | xhigh | 79.3 ±41.9 |
| 21 | Gemini 4 Argon | low | 79.2 ±16.7 |
| 22 | Kimi K3.1 | low | 79.2 ±26.6 |
| 23 | GPT-6.1 Terra Pro | medium | 79.0 ±30.0 |
| 24 | GPT-6.1 Terra | medium | 78.2 ±18.0 |
| 25 | Muse Wonder (Max) | xhigh | 77.1 ±16.9 |
| 26 | Gemini 4 Argon | medium | 76.7 ±15.6 |
| 27 | Opus 5.5 | low | 76.3 ±20.0 |
| 28 | GPT-6.1 Sol | low | 76.1 ±22.2 |
| 29 | Opus 5 | medium | 75.1 ±20.6 |
| 30 | Fable 5 | xhigh | 75.0 ±15.6 |