Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
WebGraphEvalwebgrapheval-webpage.vercel.app · Low evidenceWebGraphEval is a benchmark for evaluating web agents that aggregates multiple trajectories into a single action graph. It enables structural comparisons of agent behavior using LLM and structural signals, unlike single-path evaluation methods.
Parse Score
Sources
arxiv.org shapes more of what AI says about WebGraphEval than any other source, at 50% of its citations.
webgrapheval-webpage.vercel.app