BalatroBench - Win Rate: leaderboard
BalatroBench compares default-strategy tool-calling agents on full Balatro runs. Average final round and run win rate use all published runs, including incomplete runs. The captured v1.0.8 cohort uses the Default1.0.0 strategy, Red deck and White stake; run counts and benchmark version remain attached to each source observation. Win rate is wins divided by every published run, from 0 to 100 percent.
Metric: Run Win Rate (%). Source: balatrobench.com. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 60 | #71 |
| 2 | Gemini 3 Flash (Preview) | 33.33 | #90 |
| 3 | GPT-5.2 | 18.75 | #115 |
| 4 | Claude Sonnet 4.5 | 13.33 | #148 |
| 5 | Claude Opus 4.5 | 13.33 | #87 |
| 6 | GPT-OSS-120B | 0 | #346 |
| 7 | Claude Haiku 4.5 | 0 | #287 |
| 8 | Kimi K2.5 | 0 | #152 |
| 9 | DeepSeek V3.2 | 0 | #210 |
| 10 | Grok 4 | 0 | #189 |
| 11 | GPT-OSS-20B | 0 | #523 |
| 12 | Kimi K2 (Thinking) | 0 | #252 (Kimi K2) |
| 13 | Mistral Large 3 | 0 | #410 |
| 14 | GLM-4.7 | 0 | #202 |
| 15 | Qwen 3 Max | 0 | #213 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=balatrobench-win-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.