SEA-SpeechBench - Spoken QA (English Prompt): leaderboard
Metric: Judge score (0-100): short answers to questions about long-form Southeast Asian speech from YODAS2, rated 0-5 by a GPT-4.1 judge and multiplied by 20; English instruction prompt. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 92.28 |
| 2 | GPT-4o Audio | 86.66 |
| 3 | Qwen3 Omni 30B A3B Instruct | 85.77 |
| 4 | Voxtral-Mini-3B-2507 | 82.32 |
| 5 | Gemma 3n E4B (IT) | 79.24 |
| 6 | SeaLLMs-Audio-7B | 76.42 |
| 7 | gemma-3n-E2B-it | 73.35 |
| 8 | Qwen2.5-Omni-7B | 71.6 |
| 9 | Qwen2-Audio-7B-Instruct | 65.82 |
| 10 | Phi-4 Multimodal Instruct | 64.69 |
Interactive version: theaggregate.ai/benchmark?slug=sea-speechbench-spoken-qa-english-prompt · How It Works · Data refreshed daily, snapshot 2026-09-26.