KyrgyzLLM-Bench - KyrgyzRC: leaderboard
KyrgyzRC from KyrgyzLLM-Bench (OpenLLM Kyrgyz Leaderboard v0.1): native Kyrgyz multiple-choice reading-comprehension questions (up to five options) over short passages, zero-shot. Accuracy 0-100, higher is better; proprietary API models were run on a question subset while open-weight models used the full set.
Metric: Accuracy (%, zero-shot). Source: huggingface.co. Saturation forecast: Estimated already saturated. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 94.75 | #148 |
| 2 | GPT-5.1 (2025-11-13) | 91.5 | #107 |
| 3 | Claude Haiku 4.5 (20251001) | 91 | #264 |
| 4 | Qwen 3 Max | 91 | #213 |
| 5 | Mistral Large 3 | 90.5 | #410 |
| 6 | DeepSeek V3 | 90 | #329 |
| 7 | Gemini 2.5 Flash | 88.5 | #255 |
| 8 | Mistral Medium 3.1 | 87 | #418 |
| 9 | Grok 4.1 Fast (Non-reasoning) | 85.75 | #227 (Grok 4.1 Fast) |
| 10 | GigaChat 2 Max | 85.25 | #545 |
| 11 | GPT-5 Mini (2025-08-07) | 83.25 | #175 |
| 12 | Llama 3.1 8B Instruct | 75.2 | #1017 |
| 13 | Qwen 3 8B | 71.8 | #678 |
| 14 | Gemma 3 4B (IT) | 70.2 | #997 |
| 15 | Qwen 2.5 7B Instruct | 70 | #863 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kyrgyzllm-bench-kyrgyzrc · How It Works · Data refreshed daily, snapshot 2026-11-05.