Eye-Q - Basic - Cross-Lingual: leaderboard
Eye-Q (Najar et al., 2026), cross-lingual visual word puzzles from the Aftabe game (322 items) whose Persian answers need an English-Persian bridge of words, sounds or letters: each image encodes a hidden word or phrase through visual cues, and the model must output the exact target. The prompt gives the game rules and the answer length in characters, and the model answers once. Exact-match accuracy in percent; higher is better.
Metric: Exact-Match Accuracy (%; basic prompt with the answer length in characters, cross-lingual English-Persian subset). Source: arxiv.org. Saturation forecast: Around May 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 19.75 | #262 |
| 2 | Grok 4.1 Fast (Reasoning) | 9.72 | #426 (Grok 4.1 Fast) |
| 3 | Gemini 2.5 Flash | 4.39 | #452 |
| 4 | GPT-5.2 (Non-reasoning) | 2.82 | #172 (GPT-5.2) |
| 5 | Llama 4 Scout | 0.31 | #1787 |
| 6 | Qwen 3 VL 235B A22B Instruct | 0 | #521 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=eye-q-basic-cross-lingual · How It Works · Data refreshed daily, snapshot 2026-11-12.