GACL - Wizard: leaderboard
Game Agent Coding League: Wizard card game sub-benchmark. LLMs write code agents to play the trick-taking card game Wizard.
Metric: Normalized tournament points (0-100); source-selected best generated agent. Source: gameagentcodingleague.com. Saturation forecast: Around January 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 71.22 | #31 |
| 2 | GPT-5.4 | 66.07 | #79 |
| 3 | Gemini 3.1 Pro (Preview) | 61.77 | #58 |
| 4 | DeepSeek V4 Pro | 59.7 | #102 |
| 5 | Claude Opus 4.7 | 58.11 | #49 |
| 6 | Kimi K2.6 | 57.41 | #104 |
| 7 | Claude Opus 4.6 | 56.04 | #63 |
| 8 | Gemini 3 Flash (Preview) | 55.22 | #82 |
| 9 | Qwen 3.6 Plus | 55.15 | #120 |
| 10 | Qwen 3.6 Max Preview | 54.73 | #85 |
| 11 | MiMo-V2.5-Pro | 51.98 | #158 |
| 12 | GPT-OSS-120B | 48.69 | #341 |
| 13 | Gemma 4 31B (IT) | 47.92 | #176 |
| 14 | Claude Sonnet 4.6 | 47.61 | #90 |
| 15 | DeepSeek V4 Flash | 46.98 | #137 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gacl-wizard · How It Works · Data refreshed daily, snapshot 2026-10-20.