Benchmark
Bangla LLM Bench MMLU Bengali (285 items)
Multilingual benchmarks compare model performance across languages. Higher scores are better.
- Models measured
- 16
- Scores recorded
- 16
- Sources
- 1
- IQ weight
- 1.54
Score distribution
- 0.767
- Mean
- 0.842
- Median
- 0.157
- Std dev
- 0.642
- 10th percentile
- 0.912
- 90th percentile
- 0.393
- Min
This benchmark has a weight of 1.54 for chat models, measured over 16 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.8 Flash· Bangla LLM Bench data docs | 0.947 |
| 2 | Gemini 3.1 Pro Preview· Bangla LLM Bench data docs | 0.912 |
| 3 | Claude Opus 5.5· Bangla LLM Bench data docs | 0.905 |
| 4 | DeepSeek V4.1 Flash· Bangla LLM Bench data docs | 0.87 |
| 5 | Kimi K2.6· Bangla LLM Bench data docs | 0.87 |
| 6 | GLM 5.3· Bangla LLM Bench data docs | 0.856 |
| 7 | Claude Sonnet 5· Bangla LLM Bench data docs | 0.853 |
| 8 | Claude Haiku 4.5· Bangla LLM Bench data docs | 0.842 |
| 9 | Qwen 3.8 27B· Bangla LLM Bench data docs | 0.804 |
| 10 | GPT OSS 120B· Bangla LLM Bench data docs | 0.793 |
| 11 | Gemma 4 26B A4B· Bangla LLM Bench data docs | 0.754 |
| 12 | GPT OSS 20b· Bangla LLM Bench data docs | 0.73 |
| 13 | Qwen3 8B· Bangla LLM Bench data docs | 0.667 |
| 14 | Ministral 8B· Bangla LLM Bench data docs | 0.642 |
| 15 | Gemma 3 4B· Bangla LLM Bench data docs | 0.439 |
| 16 | Llama 3.1 8B· Bangla LLM Bench data docs | 0.393 |