HyperGVL - Neighbors: leaderboard

HyperGVL (2026) hypergraph task for vision-language models: listing the direct neighbors of a vertex, with each of 200 problems posed in 35 textual and visual representations. Accuracy (%); higher is better.

Metric: Accuracy (%) on listing the direct neighbors of a vertex, 7,000 questions per task (200 problems on synthetic and real-world hypergraphs, each posed in 35 textual, visual and combined hypergraph representations), accuracy (%) averaged over the representations, temperature 0.8, top-p 0.95; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash88.2#175
2GPT-4o83.43#751
3Gemma 3 12B73.34#2059
4Qwen 3 VL 8B55.2#917
5InternVL3-8B39.17#1782
6Qwen 2.5 VL 7B Instruct36.19#1902
7GLM-4V-9B15.46#3142
8InternVL2.5-8B10.73#2370
9Molmo-7B-D-09247.96#7207
10LLaVA-NeXT-7B6.07#5032
11DeepSeek-VL22.79#2927
12LLaVA-v1.5-7B2.64#7630

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=hypergvl-neighbors · How It Works · Data refreshed daily, snapshot 2026-11-21.