Share your thoughts, 1 month free Claude Pro on usSee more

GUI agent success on DeepShop (test)

64Success Rate

Orchard-GUI-4B (SFT + RL)

Updated 19d ago

Evaluation Results

Method	Links
Orchard-GUI-4B (SFT + RL) 2026.05		64
Gemini computer-use-preview 2026.05		62
Qwen3-VL-235B-A22B-Thinking 2026.05		56.7
Gemini-3-flash (Axtree) 2026.05		55.3
o3 (SoM) 2026.05		49.7
GPT-5 (SoM) 2026.05		49.1
Orchard-GUI-4B-SFT 2026.05		48.7
Gemini-3-flash (Axtree) 2026.05		45.1
MolmoWeb-8B 2026.05		42.3
GPT-5 (Axtree) 2026.05		40.7
MolmoWeb-4B 2026.05		35.6
Qwen3-VL-4B-Thinking 2026.05		33.3
GLM-4.1V-9B-Thinking 2026.05		32
Fara-7B 2026.05		26.2
OpenAI computer-use-preview 2026.05		24.7
GPT-4o (SoM) 2026.05		16
UI-TARS-1.5-7B 2026.05		11.6