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

GUI Reasoning on AndroidControl Low

91.8SR

UI-TARS-7B

Updated 3d ago

Evaluation Results

Method	Links
UI-TARS-7B 2026.04		91.8	-	89.3	95.2
AgentCPM-GUI-8B 2026.04		90.2	-	-	94.4
Aguvis-7B 2026.04		89.4	-	-	93.8
UI-S1-7B 2026.04		89.2	-	89.3	92.2
UI-Copilot-7B 2026.04		89.2	-	88.2	93.6
OS-Atlas-7B 2026.04		85.2	-	88	93.6
OS-Atlas-4B 2026.04		80.6	-	83.8	91.9
UI-AGILE-7B 2026.04		77.6	-	88.1	87.7
SeeClick 2026.04		75	-	73.4	93
UI-AGILE-3B 2026.04		74.3	-	87.6	85.4
OS-Genesis-7B 2026.04		74.2	-	-	90.7
UI-R1-7B 2026.04		69.7	-	83.7	86.6
NaviMaster-7B 2026.04		69.5	-	93.9	-
GUI-R1-7B 2026.04		66.5	-	85.4	85.2
UI-R1-3B 2026.04		66.4	-	82.4	79.2
GUI-R1-3B 2026.04		64.4	-	81.6	83.7
GRPO 2025.12		63.87	82.13	80.15	-
Qwen2.5VL-7B 2026.04		62.5	-	87	83.4
CAPO 2025.12		61.41	82.29	81.19	-
QwenVL2.5-3B 2025.12		59.32	62.03	74.07	-
Qwen2.5VL-3B 2026.04		59.3	-	74.1	62
Os-Atlas-4B 2025.12		40.62	64.58	71.19	-
GPT-4o 2026.04		19.4	-	0	74.3
Claude-CU 2026.04		19.4	-	0	74.3