GlobeAudio - Bengali: leaderboard

The Bengali portion of GlobeAudio (2026), a benchmark of multiple-choice questions written by native speakers about naturally occurring audio: accuracy in percent with the audio fed to the model directly. Higher is better.

Metric: Bengali (719 questions) accuracy (%) on GlobeAudio's naturalistic-audio multiple-choice questions written by native speakers; audio fed to the model directly, source-language prompts, predicted option label against the key. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)85.81#94
2Gemini 2.0 Flash68.15#704
3Gemma 3n E4B (IT)58.83#2210
4MERaLiON-2-10B56.33#1640
5MiMo-V2.555.63#354
6Voxtral-Small-24B-250745.34#1291

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=globeaudio-bengali · How It Works · Data refreshed daily, snapshot 2026-11-13.