List your company

Audio-Language Reasoning on MMAR 1.0 (test)

70.3AccuracyQwen3-Omni (T)
20.58833.49446.459.306Apr 27, 2026

Evaluation results

MethodLinks
Qwen3-Omni (T)Size=30B†, Evaluation Protocol=Full2026.0470.359.6
Qwen3-Omni (I)Size=30B†, Evaluation Protocol=Full2026.0469.763.3
Qwen2.5-Omni-7BSize=10.7B, Evaluation Protocol=Full2026.0463.964.6
Audio-Flamingo-3Size=8.4B, Evaluation Protocol=Full2026.0458.856.3
Voxtral-Mini-3BSize=4.7B, Evaluation Protocol=Full2026.0450.966.4
DeSTA-2.5Size=8.8B, Evaluation Protocol=Full2026.0446.456.2
Qwen2-Audio-7BSize=8.2B, Evaluation Protocol=Full2026.0446.356.2
Phi-4-MultimodalSize=5.6B, Evaluation Protocol=Full2026.0446.159.9
Qwen3-Omni (I)Size=30B†, Evaluation Protocol=None2026.0444.1—
Qwen3-Omni (T)Size=30B†, Evaluation Protocol=None2026.0441.9—
Qwen2.5-Omni-7BSize=10.7B, Evaluation Protocol=None2026.0441.3—
Qwen3-Omni (I)Size=30B†, Evaluation Protocol=TB2026.0437.6—
Audio-Flamingo-3Size=8.4B, Evaluation Protocol=TB2026.0435.3—
Qwen2.5-Omni-7BSize=10.7B, Evaluation Protocol=TB2026.0435.3—
Voxtral-Mini-3BSize=4.7B, Evaluation Protocol=None2026.0433.8—
Audio-Flamingo-3Size=8.4B, Evaluation Protocol=None2026.0433.1—
Qwen3-Omni (T)Size=30B†, Evaluation Protocol=TB2026.0431.6—
Phi-4-MultimodalSize=5.6B, Evaluation Protocol=TB2026.0428.3—
Phi-4-MultimodalSize=5.6B, Evaluation Protocol=None2026.0427.6—
Voxtral-Mini-3BSize=4.7B, Evaluation Protocol=TB2026.0426.3—
DeSTA-2.5Size=8.8B, Evaluation Protocol=TB2026.0426.2—
DeSTA-2.5Size=8.8B, Evaluation Protocol=None2026.0426.1—
Qwen2-Audio-7BSize=8.2B, Evaluation Protocol=None2026.0426—
Chance LevelEvaluation Protocol=Full2026.0425—
Chance LevelEvaluation Protocol=None2026.0425—
Chance LevelEvaluation Protocol=TB2026.0425—
Qwen2-Audio-7BSize=8.2B, Evaluation Protocol=TB2026.0422.5—