AgentEscapeBench - Difficulty 25: leaderboard

Metric: Success rate (%; share of the 30 difficulty-25 instances (dependency DAGs of 25 tool and item nodes) whose deterministic final flag the agent submits; single run per instance, official APIs at temperature 0.7, multi-turn tool calling with incremental disclosure of hidden nodes). Source: arxiv.org. Saturation forecast: Around 2028. 3 models tracked.

Top models

#ModelScore
1Claude Opus 4.660
2GPT-5.443.3
3Gemini 3.1 Pro (Preview)13.3

Interactive version: theaggregate.ai/benchmark?slug=agentescapebench-difficulty-25 · How It Works · Data refreshed daily, snapshot 2026-09-26.