Bengali Dialect QA Consistency - Mymensingh: leaderboard
Metric: Dialect consistency score (0-10): a Gemini 2.5 Flash judge (human review of low-confidence judgments) compares the model's answer to each human-corrected dialectal question with its answer to the standard Bengali original, rating dialect comprehension, factual correctness, completeness, clarity and length on weighted Likert scales; responses not in Bengali script or refusals score 0, Mymensingh dialect; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-OSS-120B | 9.22 | #330 |
| 2 | Gemma 3 27B (IT) | 9.16 | #509 |
| 3 | Llama 3.3 70B Instruct | 9.06 | #520 |
| 4 | Gemma 3 12B (IT) | 8.49 | #655 |
| 5 | gemma-3n-E2B-it | 7.63 | #913 |
| 6 | Gemma 3n E4B (IT) | 7.2 | #736 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=bengali-dialect-qa-consistency-mymensingh · How It Works · Data refreshed daily, snapshot 2026-10-11.