CommunityBench - Community Identification: leaderboard
Given a query, candidate replies and an observed vote distribution, the model names which Reddit community produced it. Accuracy (%), higher is better; the authors find most models keep at least 85% of this accuracy without the distribution, so it partly measures recognition of Reddit content (CommunityBench, Jiayu Lin and Zhongyu Wei, Fudan University, 2026).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 84.3 | #741 |
| 2 | DeepSeek V3 (0324) | 82.32 | #704 |
| 3 | Grok 4 | 82.23 | #306 |
| 4 | Llama 3.1 70B Instruct | 77.02 | #1541 |
| 5 | Llama 3.3 70B Instruct | 73.8 | #1353 |
| 6 | Qwen 2.5 72B Instruct | 68.96 | #1093 |
| 7 | Qwen 2.5 14B Instruct | 68.53 | #1745 |
| 8 | GLM-4 32B (0414) | 67.18 | #1407 |
| 9 | Qwen 3 32B | 60.86 | #960 |
| 10 | Qwen 2.5 7B Instruct | 58.66 | #2583 |
| 11 | glm-4-9B (0414) | 51.25 | #2196 |
| 12 | InternLM3-8B-Instruct | 49.24 | #2586 |
| 13 | Mistral 7B Instruct (v0.3) | 48.55 | #6698 |
| 14 | Qwen 3 14B | 48.52 | #1235 |
| 15 | Llama 3.1 8B Instruct | 47.37 | #4277 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=communitybench-community-identification · How It Works · Data refreshed daily, snapshot 2026-11-12.