Audio Understanding & Reasoning on MMSU
83.7ScoreGemini 3 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3 Pro2026.04 | 83.7 | |
| qwen3.5-omni-plus2026.04 | 82.74 | |
| Step-Audio-R1.52026.04 | 79.03 | |
| Gemini 3 Flash2026.04 | 76.64 | |
| Step-Audio-R12026.04 | 75.68 | |
| qwen3.5-omni-flash2026.04 | 72.5 | |
| Qwen2.5-Omni-7B + CoATBackbone=Qwen2.5-Omni-7B, Method=CoAT2026.06 | 64.99 | |
| Qwen2.5-Omni-7BBackbone=Qwen2.5-Omni-7B, Method=Baseline2026.06 | 61.32 | |
| Audio Flamingo 3Backbone=Audio Flamingo 3, Method=Baseline2026.06 | 60.01 | |
| Audio Flamingo 3 + CoATBackbone=Audio Flamingo 3, Method=CoAT2026.06 | 58.36 | |
| Qwen2-Audio + CoATBackbone=Qwen2-Audio, Method=CoAT2026.06 | 58.27 | |
| Qwen2-AudioBackbone=Qwen2-Audio, Method=Baseline2026.06 | 53.31 | |
| TACArchitecture=TAC + Text-only Reasoner, Reasoner=Gemini32026.02 | 0.724 | |
| TACArchitecture=TAC + Text-only Reasoner, Reasoner=Qwen32026.02 | 0.65 | |
| Audio Flamingo 3Architecture=Native LALM2026.02 | 0.623 |