AgentEscapeBench - Difficulty 20: leaderboard

Metric: Success rate (%; share of the 60 difficulty-20 instances (dependency DAGs of 20 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 December 2026. 11 models tracked.

Top models

#ModelScore
1Claude Opus 4.671
2Gemini 3.1 Pro (Preview)60
3GPT-5.453.3
4Kimi K2.531.7
5GPT-520
6Gemini 3 Flash (Preview)18.3
7Seed 2.0 Pro15
8MiniMax-M25

Interactive version: theaggregate.ai/benchmark?slug=agentescapebench-difficulty-20 · How It Works · Data refreshed daily, snapshot 2026-09-26.