COIN-Bench - Breadth: leaderboard
Metric: Breadth score as printed: total weight of the COIN-Tree nodes the questionnaire covers (node weights 1-3 per discussion from upvotes and views, summed when nodes merge; no fixed maximum), for questionnaires the model writes from each product's consumer discussions, matched by sentence embeddings to COIN-Tree, a weighted intent tree that GPT-4o extracted from the same discussions; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 53.62 | #105 |
| 2 | Gemini 3 Flash | 53.51 | #93 |
| 3 | GPT-5 | 53.48 | #91 |
| 4 | GPT-4.1 | 53.41 | #240 |
| 5 | DeepSeek R1 Distill Qwen 14B | 53.32 | #828 |
| 6 | Qwen 3 30B A3B | 53.2 | #488 |
| 7 | GPT-4o | 52.95 | #333 |
| 8 | Claude 3.5 Sonnet | 52.83 | #337 |
| 9 | O3 | 52.73 | #121 |
| 10 | Qwen 2.5 32B Instruct | 52.46 | #491 |
| 11 | Qwen 3 32B | 51.15 | #424 |
| 12 | Qwen 3 8B | 50.58 | #667 |
| 13 | Qwen 2.5 72B Instruct | 50.52 | #436 |
| 14 | DeepSeek R1 Distill Qwen 32B | 50.34 | #640 |
| 15 | Qwen 2.5 14B Instruct | 48.27 | #634 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=coin-bench-breadth · How It Works · Data refreshed daily, snapshot 2026-10-11.