List your company

Agent Workflow Execution on CUA-World-Long (test)

55.5Average ScoreGPT-5.4
19.128.553847.45Apr 7, 2026

Evaluation results

MethodLinks
GPT-5.4Max steps=2000, Cost cap=none2026.0455.527.5
Gemini 3 FlashMax steps=2000, Cost cap=none2026.0438.711.5
Gemini 3 FlashMax steps=500, Cost cap=$52026.0436.27.5
Kimi-K 2.5Max steps=500, Cost cap=$52026.0433.95.5
GPT-5.4Max steps=500, Cost cap=$52026.0422.73
Sonnet 4.6Max steps=500, Cost cap=$52026.0420.56