SEABED - Prosodic Ambiguity (Open-Ended): leaderboard
SEABED (Rajgarhia et al., 2026) is an audio-first reasoning benchmark on Southeast Asian speech built from openly available corpora. In this task the model resolves structurally ambiguous Indonesian sentences whose intended reading is fixed only by the prosody of the speaker; answers are open-ended and a Grok 4.5 judge marks each answer correct or incorrect, and the score is accuracy in percent, higher is better.
Metric: Accuracy (%; 504 prosodic ambiguity resolution of structurally ambiguous Indonesian sentences items; open-ended answers judged by Grok 4.5, binary correct or incorrect (graded 0 to 1 for long-form); audio at 16 kHz). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.6 Flash | 49.4 | #72 |
| 2 | Gemini 3.5 Flash | 48.5 | #98 |
| 3 | Gemini 2.5 Pro | 35.9 | #262 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=seabed-prosodic-ambiguity-open-ended · How It Works · Data refreshed daily, snapshot 2026-11-12.