HyperGVL - Degree-Specified Vertex Counting: leaderboard

HyperGVL (2026) hypergraph task for vision-language models: counting vertices with a given degree, with each of 200 problems posed in 35 textual and visual representations. Accuracy (%); higher is better.

Metric: Accuracy (%) on counting vertices with a given degree, 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 Flash77.11#175
2GPT-4o56.57#751
3Qwen 3 VL 8B52.34#917
4Gemma 3 12B51.92#2059
5InternVL3-8B27.47#1782
6Qwen 2.5 VL 7B Instruct24.63#1902
7GLM-4V-9B17.89#3142
8InternVL2.5-8B15.93#2370
9Molmo-7B-D-092413.24#7207
10LLaVA-v1.5-7B10.74#7630
11DeepSeek-VL29.81#2927
12LLaVA-NeXT-7B8.83#5032

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

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