Fibble Arena — leaderboard
Metric: Win Rate (%). Source: drchangliu.github.io. 47 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 80 |
| 2 | Gemini 3.1 Pro (Preview) | 70.59 |
| 3 | Kimi K2.5 | 55.56 |
| 4 | Gemini 2.5 Flash | 50 |
| 5 | Qwen 3 30B A3B | 40.62 |
| 6 | Gemma 4 31B | 33.33 |
| 7 | GPT-4o | 18.18 |
| 8 | Qwen 3 14B | 17.14 |
| 9 | DeepSeek V3.2 | 15.38 |
| 10 | Qwen 3 8B | 14.29 |
| 11 | Gemma 3 27B | 11.11 |
| 12 | Qwen 3.5 122B A10B | 9.68 |
| 13 | Gemma 4 26B | 8.33 |
| 14 | DeepSeek V3 | 8 |
| 15 | GPT-5 | 7.41 |
Interactive version: theaggregate.ai/benchmark?slug=fibble-arena · How the rankings work · Data refreshed daily, snapshot 2026-07-22.