ArgBench (Chain-of-Thought) - Argument Mining: leaderboard
ArgBench (2026) unifies 46 computational-argumentation tasks from 33 existing datasets into five skills. This board is the macro F1 averaged over the argument mining tasks under chain-of-thought prompting, times 100; higher is better.
Metric: Macro F1 (0-1, times 100) averaged over the argument mining (unit segmentation and classification) tasks, chain-of-thought prompting; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 | 54.2 | #506 |
| 2 | Llama 3.3 70B Instruct | 46 | #1394 |
| 3 | Qwen 3 32B | 45 | #983 |
| 4 | Mistral 7B Instruct (v0.3) | 44.1 | #6759 |
| 5 | Mixtral 8x7B Instruct | 40.9 | #4692 |
| 6 | DeepSeek R1 Distill Qwen 32B | 40.6 | #1844 |
| 7 | Mistral-Small-Instruct-2409 | 40 | #2026 |
| 8 | Phi-mini-MoE-instruct | 39.9 | |
| 9 | Llama 3.1 8B Instruct | 39.6 | #4655 |
| 10 | Qwen 3 4B | 37.5 | #2379 |
| 11 | DeepSeek-R1-Distill-Qwen-7B | 37.3 | #7976 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=argbench-chain-of-thought-argument-mining · How It Works · Data refreshed daily, snapshot 2026-11-22.