MemeBench: leaderboard
Metric: Success rate (%; visual, identity, knowledge and reasoning checklists all passed) on 1,253 Chinese and English memes centred on anime, comics, games and online subcultures, explained closed-book from the image; scored against VIKR checklists under the intersection of Gemini-3.1-Pro and GPT-5.1 judges, unscored responses counted as failures. Source: arxiv.org. Saturation forecast: Around March 2027. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 60.3 |
| 2 | Gemini 3 Flash | 53.7 |
| 3 | GPT-5.6 Sol | 40.9 |
| 4 | Claude Opus 5 | 38.3 |
| 5 | Kimi K2.5 | 37.9 |
| 6 | Qwen 3.6 Plus | 28.4 |
| 7 | GPT-5.6 Terra | 27.8 |
| 8 | GPT-5.6 Luna | 18 |
| 9 | Claude Sonnet 5 | 17.8 |
| 10 | Qwen 3 VL 235B A22B Instruct | 14.5 |
| 11 | MiMo-V2-Omni | 14.3 |
| 12 | MiMo-V2.5 | 13.7 |
| 13 | Qwen 3.6 Flash | 12.3 |
| 14 | Qwen 3 VL 235B A22B (Thinking) | 12 |
| 15 | Qwen 3 VL 32B Instruct | 8.4 |
Interactive version: theaggregate.ai/benchmark?slug=memebench · How It Works · Data refreshed daily, snapshot 2026-09-29.