GABench - Graph Retrieval - Graph Level (Numerical-Attribute Graphs): leaderboard
Metric: Task success rate (%; share of tasks whose final answer a DeepSeek-V4-Flash judge marks consistent with the verified ground truth; graph-level graph retrieval (node attributes, neighbourhoods, edges, subgraphs and graph statistics) with its 10 retrieval tools on numerical-attribute graphs (DGraph-Fin, Pokec, NBA, Credit and Bail, 200 generated tasks per dataset and category); LLM agent in the OpenClaw harness, one run). Source: arxiv.org. Saturation forecast: Around 2030. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Pro | 8.27 |
| 2 | Gemini 3 Flash | 5.34 |
| 3 | GLM-5-Turbo | 5.34 |
| 4 | Qwen 3 235B A22B | 5 |
| 5 | GPT-5 Mini | 4 |
| 6 | GLM-4.5 Flash | 1.62 |
Interactive version: theaggregate.ai/benchmark?slug=gabench-graph-retrieval-graph-level-numerical-attribute-graphs · How It Works · Data refreshed daily, snapshot 2026-09-29.