WebForge-Bench - Consumer Transaction: leaderboard
Metric: Accuracy (%) on the consumer transaction and service tasks of the 934 WebForge-Bench tasks on generated, self-contained websites with injected real-web noise, up to 50 browser actions in a Chromium GUI, final-state answer checking; multimodal setting: screenshot plus DOM; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 72.2 |
| 2 | Gemini 3 Flash | 65.2 |
| 3 | Kimi K2.5 | 60 |
| 4 | Claude Sonnet 4.5 | 58.3 |
| 5 | GPT-5 Mini | 51.3 |
| 6 | GPT-5.2 | 48.7 |
| 7 | Qwen 3 VL 235B A22B | 37.4 |
| 8 | Gemini 2.5 Flash Lite | 34.8 |
| 9 | GPT-5 Nano | 20.9 |
| 10 | qwen3-omni-30B | 6.1 |
Interactive version: theaggregate.ai/benchmark?slug=webforge-bench-consumer-transaction · How It Works · Data refreshed daily, snapshot 2026-10-07.