HyperGVL - Order-Weighted Maximum Flow: leaderboard

HyperGVL (2026) hypergraph task for vision-language models: the maximum flow between two vertices with hyperedge order as capacity, with each of 200 problems posed in 35 textual and visual representations. Accuracy (%); higher is better.

Metric: Accuracy (%) on the maximum flow between two vertices with hyperedge order as capacity, 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 Flash76.29#180
2GPT-4o26.29#754
3Qwen 3 VL 8B25.09#927
4Gemma 3 12B22.71#2091
5Qwen 2.5 VL 7B Instruct14.53#1919
6InternVL3-8B11.69#1795
7GLM-4V-9B8.89#3164
8Molmo-7B-D-09246.74#7206
9InternVL2.5-8B6.7#2393
10LLaVA-v1.5-7B4.51#7642
11LLaVA-NeXT-7B3.84#5045
12DeepSeek-VL23.07#3056

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=hypergvl-order-weighted-maximum-flow · How It Works · Data refreshed daily, snapshot 2026-11-22.