V-ICAL Bench - Pass Rate: leaderboard
V-ICAL Bench's second metric: the percentage of its 342 video-demonstration game and control tasks in which the agent meets the task's predefined success condition (reaching a goal, triggering a required event, hitting a target score or surviving the horizon), higher is better. The condition is judged independently of the normalized progress score; agents learn each task from one to four human demonstration videos and act from raw frames with rewards hidden (Fan et al., 2026).
Metric: Pass rate (%). Source: arxiv.org. Saturation forecast: Around February 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Seed 2.1 Pro | 58 | #133 |
| 2 | Gemini 3.1 Pro (Preview) | 53 | #91 |
| 3 | Gemini 3.6 Flash | 53 | #72 |
| 4 | GPT-5.6 Sol | 51 | #29 |
| 5 | Gemini 3 Flash | 42 | #159 |
| 6 | Seed 2.0 Lite | 41 | #286 |
| 7 | Kimi K3 | 39 | #53 |
| 8 | DeepSeek V4.1 Flash | 35 | #58 |
| 9 | Qwen 3.5 397B A17B | 31 | #242 |
| 10 | Qwen 3.6 27B | 27 | #339 |
| 11 | Gemma 4 31B (IT) | 25 | #343 |
| 12 | Qwen 3.5 27B | 24 | #358 |
| 13 | Qwen 2.5 VL 72B Instruct | 23 | #845 |
| 14 | MiMo-V2.5 | 23 | #353 |
| 15 | GLM-4.5V | 21 | #739 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=v-ical-bench-pass-rate · How It Works · Data refreshed daily, snapshot 2026-11-12.