Exgentic Open Agent - AppWorld: leaderboard

Exgentic Open Agent AppWorld evaluates agent frameworks and models on simulated app-control tasks with API/tool use.

Metric: Success Rate (%). Source: www.exgentic.ai. 25 models tracked.

Top models

#ModelScoreOverall rank
1Smolagent + claude-opus-4.570
2OpenAI Solo + claude-opus-4.568
3Claude Code + claude-opus-4.566
4React + Shortlisting + claude-opus-4.564
5React + claude-opus-4.561
6OpenAI Solo + gemini-3-pro57
7React + Shortlisting + gemini-3-pro55
8React + gemini-3-pro50
9Claude Code + gemini-3-pro36
10React + Shortlisting + gpt-5.222

Interactive version: theaggregate.ai/benchmark?slug=exgentic-open-agent-appworld · How It Works · Data refreshed daily, snapshot 2026-10-20.