ArmBench-LLM - Text Processing: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 Pro | 87.91 |
| 2 | Gemini 2.5 Flash | 85.63 |
| 3 | Gemma 4 31B (IT) | 83.69 |
| 4 | Gemini 3 Flash (Preview) | 83.57 |
| 5 | GPT-5.4 Mini | 81.38 |
| 6 | Mistral Large 3 | 81.17 |
| 7 | Claude 3.7 Sonnet | 80.92 |
| 8 | Grok 4 Fast | 77.88 |
| 9 | Claude Sonnet 4 | 76.77 |
| 10 | Grok 4.1 Fast | 76.61 |
| 11 | Mistral Small 3.2 | 76.5 |
| 12 | GPT-5.4 Nano | 76.31 |
| 13 | Gemma 3 27B (IT) | 76.17 |
| 14 | Gemini 3 Pro (Preview) | 76.02 |
| 15 | Qwen 3.5 27B | 73.2 |
Interactive version: theaggregate.ai/benchmark?slug=armbench-llm-text-processing · How It Works · Data refreshed daily, snapshot 2026-09-19.