ArgBench (Chain-of-Thought) - Perspective Assessment: leaderboard
ArgBench (2026) unifies 46 computational-argumentation tasks from 33 existing datasets into five skills. This board is the macro F1 averaged over the perspective assessment tasks under chain-of-thought prompting, times 100; higher is better.
Metric: Macro F1 (0-1, times 100) averaged over the argument perspective assessment (stance, framing and similar) 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 | 60.3 | #505 |
| 2 | Qwen 3 32B | 55.2 | #972 |
| 3 | Llama 3.3 70B Instruct | 53.1 | #1385 |
| 4 | Mixtral 8x7B Instruct | 52.3 | #4696 |
| 5 | Llama 3.1 8B Instruct | 49.2 | #4600 |
| 6 | Qwen 3 4B | 48.4 | #2348 |
| 7 | DeepSeek R1 Distill Qwen 32B | 47.8 | #1834 |
| 8 | Mistral-Small-Instruct-2409 | 47.4 | #2016 |
| 9 | DeepSeek-R1-Distill-Qwen-7B | 42.2 | #8055 |
| 10 | Phi-mini-MoE-instruct | 39.8 | |
| 11 | Mistral 7B Instruct (v0.3) | 38.8 | #6750 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=argbench-chain-of-thought-perspective-assessment · How It Works · Data refreshed daily, snapshot 2026-11-21.