R2-Omni - DailyOmni (Uncompressed): leaderboard
R2-Omni v1 AppendixA Table6, Qwen2.5-Omni-3B original BF16 model with no token pruning, greedy decoding, DailyOmni multiple-choice accuracy. AppendixA.1 explicitly binds this experiment to the3B model; other pruning fractions are separate interventions. Reported metric: Multiple-choice accuracy (%).
Metric: Multiple-choice accuracy (%). Source: arxiv.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen2.5-Omni-3B | 55.56 |
Interactive version: theaggregate.ai/benchmark?slug=r2-omni-dailyomni-uncompressed · How It Works · Data refreshed daily, snapshot 2026-11-05.