ArmBench-LLM - Exams: leaderboard
ArmBench-LLM category: Armenian Unified Exams in mathematics, literature and history, each graded on the exams' 0-20 points scale; the score is the mean of the three exams, 0-20, higher is better.
Metric: Exam score (0-20). Source: huggingface.co. Saturation forecast: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 14.17 | #90 |
| 2 | Gemini 3 Pro (Preview) | 13.92 | #71 |
| 3 | GPT-5.2 Pro | 13.75 | #70 |
| 4 | Gemini 2.5 Pro | 12.75 | #159 |
| 5 | Grok 4.1 Fast | 11.58 | #227 |
| 6 | Gemini 3.1 Pro (Preview) | 11.33 | #64 |
| 7 | Grok 4 Fast | 11.33 | #266 |
| 8 | Qwen 3.5 27B | 10.83 | #192 |
| 9 | Gemini 2.5 Flash | 10.75 | #255 |
| 10 | Claude 3.7 Sonnet | 10.67 | #259 |
| 11 | Mistral Large 3 | 10.33 | #410 |
| 12 | Claude Sonnet 4 | 10.17 | #207 |
| 13 | Gemma 4 31B (IT) | 10 | #178 |
| 14 | GLM-5 | 9 | #149 |
| 15 | Qwen 3.5 35B A3B | 8.5 | #268 |
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-exams · How It Works · Data refreshed daily, snapshot 2026-11-05.