ArgBench (Few-Shot) - Quality 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 argument quality assessment tasks under few-shot prompting, times 100; higher is better.

Metric: Macro F1 (0-1, times 100) averaged over the argument quality assessment tasks, few-shot prompting; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-4.147.7#505
2Qwen 3 32B42.2#972
3Llama 3.3 70B Instruct41.8#1385
4DeepSeek R1 Distill Qwen 32B40.2#1834
5DeepSeek-R1-Distill-Qwen-7B39.5#8055
6Mixtral 8x7B Instruct39#4696
7Qwen 3 4B36.7#2348
8Llama 3.1 8B Instruct35.8#4600
9Mistral 7B Instruct (v0.3)34.4#6750
10Phi-mini-MoE-instruct32.9
11Mistral-Small-Instruct-240931.1#2016

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=argbench-few-shot-quality-assessment · How It Works · Data refreshed daily, snapshot 2026-11-21.