Konkur 1404 - Humanities: leaderboard
The Humanities stream of Konkur 1404 — Persian-language multiple-choice questions from the humanities exam of Iran's 1404 (2025) national university entrance exam; text-only accuracy (%).
Metric: Accuracy (%, text-only). Source: huggingface.co. Saturation forecast: Around September 2028. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B Instruct | 52.15 | #273 |
| 2 | Qwen 3 Next 80B A3B Instruct | 46.24 | #368 |
| 3 | Devstral Small 2 | 45.7 | #495 |
| 4 | GPT-OSS-120B | 45.16 | #341 |
| 5 | Kimi K2 0905 | 45.16 | #288 |
| 6 | gpt-oss-safeguard-20B | 44.62 | #314 |
| 7 | Llama 4 Maverick | 43.55 | #464 |
| 8 | Llama 3.1 70B Instruct | 43.01 | #561 |
| 9 | MiMo-V2-Flash | 42.47 | #372 |
| 10 | Qwen 3 VL 30B A3B Instruct | 41.4 | #384 |
| 11 | Gemma 3 27B (IT) | 40.32 | #523 |
| 12 | DeepSeek V3.2 | 40.32 | #205 |
| 13 | Ministral 3 14B | 38.17 | #650 |
| 14 | Mixtral 8x22B Instruct | 37.1 | #728 |
| 15 | Gemma 3n E4B (IT) | 35.48 | #749 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=konkur-1404-humanities · How It Works · Data refreshed daily, snapshot 2026-10-20.