Omni2Web - Instruction Utility: leaderboard
Omni2Web (Han et al., 2026) measures how useful the instructions recovered by each model are: a fixed DeepSeek-v4-pro coding model executes them on the source page, and the result is scored with the direct-editing rubric. Edit Fidelity Score in percent over the 918 recordings; higher is better.
Metric: Edit Fidelity Score (%; Track C: the model's Track B instructions are executed on the source HTML by a fixed DeepSeek-v4-pro coding model, then scored with the Track A rubric: 0.10 grounding + 0.50 edit fulfillment + 0.20 no-damage + 0.20 revision handling; 918 instances; judged by Gemini 3.1 Pro). Source: arxiv.org. Saturation forecast: Not forecast. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 Omni Plus | 51.08 | #237 |
| 2 | Gemini 3.5 Flash | 49.77 | #98 |
| 3 | Seed 2.0 Lite | 47.42 | #286 |
| 4 | Claude Opus 5 | 47.07 | #20 |
| 5 | GPT-5.6 Sol | 46.19 | #29 |
| 6 | Gemini 3.1 Pro (Preview) | 45.53 | #91 |
| 7 | Kimi K3 | 40.91 | #53 |
| 8 | Qwen 3.8 Max | 39.52 | #55 |
| 9 | Seed 2.1 Pro | 38.66 | #133 |
| 10 | Grok 4.6 | 38.6 | #44 |
| 11 | MiMo-V2.5 | 37.87 | #353 |
| 12 | MiniMax-M3 | 28.67 | #265 |
| 13 | Qwen 3.8 27B | 28.48 | #199 |
| 14 | Muse Spark 1.1 | 24.37 | #82 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=omni2web-instruction-utility · How It Works · Data refreshed daily, snapshot 2026-11-12.