SOTA GUI Grounding on MMBench-GUI-L2 and PapersWithCode

91.3Accuracy

MAI-UI-32B

Updated 24d ago

Evaluation Results

Method	Links
MAI-UI-32B 2026.06		91.3
MAI-UI-8B 2026.06		88.8
UI-Venus-1.5-30B-A3B 2026.06		88.6
UI-Venus-1.5-8B 2026.06		88.1
GUI-Owl-1.5-32B-Instruct 2026.06		86.8
GUI-SD 2026.05		86.7
UI-Venus-Ground-72B 2026.06		86.3
Step-GUI-8B 2025.12		85.6
Step-GUI-8B 2026.06		85.6
GUI-Owl-1.5-32B-Thinking 2026.06		84.5
SeedVL-1.5 2025.12		84.4
Qwen3-VL-Instruct + GRPO-Binary 2026.05		84.3
Step-GUI-4B 2025.12		84
Step-GUI-4B 2026.06		84
Qwen3-VL-Instruct + GRPO-Gaussian 2026.05		83.7
Qwen3-VL-30B-A3B 2026.06		83.7
Qwen3-VL-Instruct + GRPO-Distance 2026.05		83.3
GUI-Owl-1.5-4B-Instruct 2026.06		83.2
UI-Ins-7B 2026.05		83.1
GUI-Owl-32B 2025.12		83
Qwen3-VL-Instruct 2026.05		83
GUI-Owl-32B 2026.06		83
Xiaomi-GUI-0-30B-A3B 2026.06		82.7
GUI-Owl-1.5-8B-Instruct 2026.06		82.5
Qwen3-VL-8B 2026.06		81.4
GUI-Owl-7B 2025.12		80.5
GUI-Owl-7B 2026.06		80.5
Qwen3-VL-8B 2026.04		80.1
GUI-Owl-1.5-8B-Thinking 2026.06		80.1
UI-Venus-Ground-7B 2026.06		79.9
Qwen3-30B-A3B-Instruct 2025.12		79.7
HyperClick-7B 2026.05		79.6
Ours (Qwen3-VL-8B) 2026.04		78.4
Propose-then-Critic-8B 2026.05		78.4
Qwen3-VL-4B 2026.04		77.4
UI-TARS-72B 2026.06		74.3
UI-TARS-72B-DPO 2025.12		74.2
ZoomUI-7B 2026.05		72.8
Ours (UI-Tars1.5) 2026.04		68.9
UGround-V1-7B 2026.04		65.6
UI-TARS-1.5-7B 2025.12		64.3
UI-TARS-1.5-7B 2026.04		64.3
UI-TARS-1.5-7B 2026.06		64.3
GRPO w. Two. 2025.11		60.6
GRPO w. Max. 2025.11		58.2
GRPO w. Min. 2025.11		58
GRPO w. Two. 2025.11		57.4
GRPO 2025.11		57
GRPO w. Min. 2025.11		55.6
GRPO w. Max. 2025.11		54.6
GRPO 2025.11		53.8
GRPO w. Min. 2025.11		51
GRPO w. Two. 2025.11		49
GRPO w. Max. 2025.11		48.8
Qwen3-VL-2B 2026.04		48.1
GRPO 2025.11		46.4
Base Model 2025.11		45
Qwen2.5-VL-72B 2025.12		41.8
OS-Atlas-Base-7B 2026.04		41.4
Qwen2.5-VL-7B 2026.04		33.8
Qwen2.5-VL-3B 2026.04		21.2
Claude-3.7 Sonnet 2026.04		4.66
GPT-4o 2026.04		2.87