DedeuceBench — leaderboard

LLM-agent active-learning benchmark with budgeted search results, trap-free rate, query usage, and token-count telemetry.

Metric: Score100. Source: huggingface.co. 2 models tracked.

Top models

#ModelScore
1openai:gpt-5-mini95.31
2openai:gpt-5-nano50

Interactive version: theaggregate.ai/benchmark?slug=dedeucebench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.