MCGA - Speech Reasoning: leaderboard
Metric: Accuracy (%; multiple-choice questions on 1,000 test recordings that need external knowledge). Source: arxiv.org. Saturation forecast: Around December 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 83.7 |
| 2 | Qwen3 Omni 30B A3B Instruct | 82.9 |
| 3 | Qwen2-Audio-7B-Instruct | 64.7 |
| 4 | Phi-4 Multimodal Instruct | 54.4 |
Interactive version: theaggregate.ai/benchmark?slug=mcga-speech-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-25.