EgoIntent: leaderboard
Metric: Overall semantic agreement, the mean of the Local Intent, Procedural Intent and Next-Plan scores (0-100) from a DeepSeek-V4-Flash judge comparing each open-ended prediction with the human reference over the 3,014 EgoIntent micro-steps (pre-outcome visual prefix only, model-specific video or multi-frame interface; missing predictions score 0); higher is better. Source: arxiv.org. Saturation forecast: Around July 2028. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Seed 2.1 Turbo | 55.13 | #84 |
| 2 | Qwen 3.5 Plus | 52.71 | #123 |
| 3 | Qwen 3 VL 32B Instruct | 42.8 | #276 |
| 4 | Gemini 3.5 Flash | 41.45 | #55 |
| 5 | Qwen 3 VL 8B Instruct | 37.49 | #401 |
| 6 | Nova 2.0 Lite | 34.59 | #408 |
| 7 | Qwen 2.5 VL 7B Instruct | 33.46 | #643 |
| 8 | Qwen 2 VL 7B Instruct | 33.18 | #816 |
| 9 | Molmo2-8B | 32.22 | #522 |
| 10 | InternVL3-8B | 30.97 | #606 |
| 11 | InternVL2-8B | 20.81 | #826 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=egointent · How It Works · Data refreshed daily, snapshot 2026-10-11.