Beyond Binary: leaderboard
Test-Time Scaling (TTS), which samples multiple candidate actions and ranks them via a Critic Model, has emerged as a promising paradigm for generalist GUI agents. Its efficacy thus hinges on the critic's fine-grained ranking ability. However, existing GUI critic models uniformly adopt binary classification. Our motivational analysis of these models exposes a severe entanglement: scores for valid
Metric: NDCG@All (self-reported). Source: benchmarklist.com. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 84.52 | #98 |
| 2 | GPT-4o | 80.68 | #343 |
| 3 | Qwen 2.5 VL 7B | 77.78 | #643 |
| 4 | Claude Sonnet 4 | 76.75 | #199 |
| 5 | Gemini 3 Pro | 72.62 | #80 |
| 6 | uitars-1.5-7B | 56.78 | #864 |
| 7 | Gemini 3 Flash | 55.5 | #99 |
Interactive version: theaggregate.ai/benchmark?slug=beyond-binary · How It Works · Data refreshed daily, snapshot 2026-10-20.