Done, But Not Sure: leaderboard
Standard embodied evaluations do not independently score whether an agent correctly commits to task completion at episode closure, a capacity we call terminal commitment. Behaviorally distinct failures--never completing the task, completing it but failing to stop, and reporting success without sufficient evidence--collapse into the same benchmark failure. We introduce VIGIL, an evaluation framewor
Metric: B All (self-reported). Source: benchmarklist.com. Saturation forecast: Around August 2027. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 56.4 | #58 |
| 2 | Qwen 3.6 27B | 37.8 | #170 |
| 3 | Seed 1.8 | 37.1 | #139 |
| 4 | Qwen 3.5 27B | 36.2 | #187 |
| 5 | GPT-5.4 | 24.7 | #79 |
| 6 | Claude Sonnet 4 | 21.1 | #199 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=done-but-not-sure · How It Works · Data refreshed daily, snapshot 2026-10-20.