New metric for agent trajectories moves beyond binary success flags
Researchers introduce OTAP, a novel evaluation framework that moves beyond binary success metrics for agent trajectories by modeling execution as a graph distance problem. Rather than penalizing valid but structurally different solutions, the method uses optimal transport over dependency graphs to measure semantic equivalence between an agent's plan and reference solutions. This addresses a critical gap in LLM agent benchmarking, where current metrics either obscure failure modes or unfairly reject correct reasoning paths that diverge from canonical forms. The work matters for anyone building or evaluating agentic systems, as it enables more nuanced assessment of planning quality and execution robustness.
Modelwire context
ExplainerThe deeper issue OTAP is solving is not just metric granularity but a structural flaw in how agent benchmarks are constructed: most assume a single canonical execution path, which means a correct agent that solves a task differently gets penalized the same as one that fails outright. OTAP reframes evaluation as a graph alignment problem, borrowing Gromov-Wasserstein transport from geometric deep learning rather than building on existing NLP similarity approaches.
This is largely disconnected from recent activity in our archive, as we have no prior coverage of agent evaluation methodology to anchor it to. It belongs to a growing body of work questioning whether current agentic benchmarks measure what they claim to measure. The concern is not new in the research community, but tooling that operationalizes a better metric is rarer than papers that simply critique existing ones. OTAP is notable precisely because it ships a concrete scoring mechanism rather than stopping at diagnosis.
The real test is whether OTAP gets adopted by any of the major agent benchmark maintainers (WebArena, SWE-bench, or similar) within the next six to twelve months. Adoption there would signal the field is ready to move past binary pass-fail scoring; continued absence would suggest the computational overhead of optimal transport is a practical barrier.
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsOTAP · Gromov-Wasserstein transport · LLM agents
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. arXiv cs.LG originally reported this story as “Otap:Structure-Aware Optimal Transport for Evaluating Planning and Execution in Agent Trajectories”. The full content lives on arxiv.org. If you’re a publisher and want a different summarization policy for your work, see our takedown page.