ERQA: leaderboard
Embodied Reasoning Question Answering benchmark with visual questions about space, trajectories, actions, state estimation, and multi-view reasoning.
Metric: Score (%). Source: github.com. Source: llm-stats.com. Saturation forecast: Around October 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.8 Max | 77.8 | #35 |
| 2 | Qwen3.8-Flash-Next | 72.3 | #72 |
| 3 | Seed 2.1 Pro | 72 | #67 |
| 4 | Seed 2.1 Turbo | 71.3 | #86 |
| 5 | Qwen 3.7 Plus | 69.8 | #92 |
| 6 | GPT-5 | 65.7 | #98 |
| 7 | Qwen 3.6 Plus | 65.7 | #120 |
| 8 | Qwen 3.8 27B | 65.5 | #105 |
| 9 | Qwen 3.5 35B A3B | 64.8 | #256 |
| 10 | Muse Spark | 64.7 | #64 |
| 11 | O3 | 64 | #124 |
| 12 | Qwen 3.6 27B | 62.5 | #170 |
| 13 | Qwen 3.5 122B A10B | 62 | #173 |
| 14 | Qwen 3.5 27B | 60.5 | #187 |
| 15 | Seed 1.8 | 58.8 | #139 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=erqa · How It Works · Data refreshed daily, snapshot 2026-10-20.