ArmBench-LLM - Classification: leaderboard
ArmBench-LLM category: Armenian text classification, the mean accuracy on the Topic-14 topic classification and sentiment analysis tasks. 0-100, higher is better.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around 2030. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5 | 64.57 | #149 |
| 2 | Claude 3.7 Sonnet | 64.43 | #259 |
| 3 | Gemma 4 31B (IT) | 64.29 | #178 |
| 4 | Gemini 3 Pro (Preview) | 64.14 | #71 |
| 5 | Gemini 3.1 Pro (Preview) | 63.96 | #64 |
| 6 | Gemini 2.5 Pro | 63.86 | #159 |
| 7 | Gemini 3 Flash (Preview) | 63.79 | #90 |
| 8 | GPT-5.4 Nano | 62.14 | #359 |
| 9 | Qwen 3.5 27B | 61.21 | #192 |
| 10 | GLM-4.7 Flash | 59.86 | #543 |
| 11 | Claude Sonnet 4 | 59.82 | #207 |
| 12 | Gemma 3 12B (IT) | 59.71 | #672 |
| 13 | Gemini 2.5 Flash | 59.54 | #255 |
| 14 | Qwen 3.5 35B A3B | 59.5 | #268 |
| 15 | Ministral 3 14B | 59.07 | #664 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=armbench-llm-classification · How It Works · Data refreshed daily, snapshot 2026-11-05.