ArgBench (Zero-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 zero-shot prompting, times 100; higher is better.

Metric: Macro F1 (0-1, times 100) averaged over the argument quality assessment tasks, zero-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-4.144#505
2Llama 3.3 70B Instruct41.9#1385
3Qwen 3 32B39.4#972
4DeepSeek-R1-Distill-Qwen-7B38.3#8055
5Llama 3.1 8B Instruct37.7#4600
6DeepSeek R1 Distill Qwen 32B37.3#1834
7Mistral 7B Instruct (v0.3)36#6750
8Mistral-Small-Instruct-240933.5#2016
9Mixtral 8x7B Instruct33.2#4696
10Qwen 3 4B33.2#2348
11Phi-mini-MoE-instruct32.1

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

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