AMIGO (Threshold 0.3): leaderboard
Metric: Verified Correct accuracy in percent on AMIGO's dress galleries at similarity threshold 0.3 (770 galleries): the model asks Yes/No attribute questions (at most 20 after upload) to a Qwen3-VL-235B oracle and its final guess is correct and the dialogue leaves the target as the unique best-supported candidate; higher is better. Source: arxiv.org. Saturation forecast: Around December 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B (Non-reasoning) | 25.2 | #195 (Gemma 4 31B) |
| 2 | Qwen 3.5 397B A17B (Non-reasoning) | 7.4 | #141 (Qwen 3.5 397B A17B) |
| 3 | Gemma 4 12B (Non-reasoning) | 4.9 | #529 (Gemma 4 12B) |
| 4 | Step3 VL 10B | 1.3 | #465 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=amigo-threshold-0-3 · How It Works · Data refreshed daily, snapshot 2026-10-11.