ArgBench (Chain-of-Thought): leaderboard
Metric: Macro average (0-1 F1-type scores, times 100) over the five skills (mining, perspective, quality, reasoning, and generation scored by BERTScore) of 46 tasks from 33 existing argumentation datasets, chain-of-thought prompting; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 57.7 |
| 2 | Llama 3.3 70B Instruct | 52.9 |
| 3 | Qwen 3 32B | 52.2 |
| 4 | Mixtral 8x7B Instruct (v0.1) | 50.1 |
| 5 | Mistral-Small-Instruct-2409 | 49.7 |
| 6 | Llama 3.1 8B Instruct | 49.4 |
| 7 | Mistral 7B Instruct (v0.3) | 48.3 |
| 8 | DeepSeek R1 Distill Qwen 32B | 47.7 |
| 9 | Qwen 3 4B | 47.1 |
| 10 | DeepSeek-R1-Distill-Qwen-7B | 45.4 |
Interactive version: theaggregate.ai/benchmark?slug=argbench-chain-of-thought · How It Works · Data refreshed daily, snapshot 2026-10-07.