Visual Intelligence Leaderboard

Visual Intelligence Leaderboard — Benchmark Data A held-out benchmark for multimodal LLMs, spanning two tracks of visual intelligence: Track 1 · Do You See Me (low-level visual perception) and Track 2 · Mind’s Eye (visuo-cognitive reasoning). ⚠️ Held-out benchmark. This repository ships the questions and images only. The ground-truth answers are withheld to keep the leaderboard fair and un-gameable. See Evaluation below for how to have a model scored. Subsets… See the full description on the dataset page:

Source: Hugging Face Hub (amolharsh/visual-intelligence-leaderboard). Metadata imported from the dataset’s Hub tags.