ArgBench (Zero-Shot) - Argument Reasoning: leaderboard
Metric: Macro F1 (0-1, times 100) averaged over the argument reasoning (fallacies, warrants and similar) tasks, zero-shot prompting; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 55.6 |
| 2 | Llama 3.3 70B Instruct | 52.4 |
| 3 | Qwen 3 32B | 48.4 |
| 4 | Mixtral 8x7B Instruct (v0.1) | 39.5 |
| 5 | Llama 3.1 8B Instruct | 38.7 |
| 6 | Mistral 7B Instruct (v0.3) | 35.6 |
| 7 | Mistral-Small-Instruct-2409 | 34.5 |
| 8 | Qwen 3 4B | 33.6 |
| 9 | DeepSeek R1 Distill Qwen 32B | 29.2 |
| 10 | DeepSeek-R1-Distill-Qwen-7B | 22.6 |
Interactive version: theaggregate.ai/benchmark?slug=argbench-zero-shot-argument-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-07.