Mobile GUI Navigation on MobileWorld GUI-Only 117 tasks (test)
43.9Pass@1 Success RateGUI-Owl-1.5-32B-Instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| GUI-Owl-1.5-32B-InstructMemory Type=Action-Thought2026.06 | 43.9 | |
| GUI-Owl-1.5-32BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 43.9 | |
| ForgeOwl-8BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only, Adaptation Data=AndroidWorld-derived adaptation data only2026.06 | 41 | |
| MAI-UI-235B-A22BMemory Type=Action-Thought2026.06 | 39.7 | |
| MAI-UI-235B-A22BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 39.7 | |
| GUI-Owl-1.5-8B-InstructMemory Type=Action-Thought2026.06 | 38.2 | |
| GUI-Owl-1.5-8BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 37.6 | |
| MAI-UI-32BMemory Type=Action-Thought2026.06 | 36.2 | |
| MAI-UI-32BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 36.2 | |
| GUI-Owl-1.5-2B-InstructMemory Type=Action-Thought2026.06 | 32.2 | |
| GUI-Owl-1.5-2BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 32.2 | |
| MemGUI-Agent-235BMemory Type=ConAct, Evaluation Protocol=zero-shot2026.06 | 29.1 | |
| MAI-UI-8BMemory Type=Action-Thought2026.06 | 27.5 | |
| MAI-UI-8BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 27.5 | |
| MemGUI-8B-SFTMemory Type=ConAct, Evaluation Protocol=SFT2026.06 | 17.9 | |
| OpenMobile-8BMemory Type=Action-Thought2026.06 | 17.7 | |
| OpenMobile-8BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only2026.06 | 17.7 | |
| ClawGUI-2BMemory Type=Action-Thought2026.06 | 17.1 | |
| ClawGUI-2BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only2026.06 | 17.1 | |
| OpenMobile-7BMemory Type=Action-Thought2026.06 | 14.8 | |
| OpenMobile-7BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only2026.06 | 14.8 | |
| Qwen3-VL-235B-ThinkingMemory Type=Action-Thought2026.06 | 14.5 | |
| Qwen3-VL-235B-ThinkingAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 14.5 | |
| Qwen3-VL-235B-InstructMemory Type=Action-Thought2026.06 | 12.8 | |
| Qwen3-VL-235BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 12.8 | |
| Qwen3-VL-32B-InstructMemory Type=Action-Thought2026.06 | 11.9 | |
| Qwen3-VL-32BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 11.9 | |
| ForgeQwen3-8BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only, Adaptation Data=AndroidWorld-derived adaptation data only2026.06 | 10.3 | |
| Qwen3-VL-8B-InstructMemory Type=Action-Thought2026.06 | 9.4 | |
| ScaleCUA-7BMemory Type=Action-Thought2026.06 | 7.7 | |
| ScaleCUA-7BAgent Category=Open-data mobile GUI agents, Reporting Protocol=GUI-only2026.06 | 7.7 | |
| Qwen3-VL-8BAgent Category=Open-weight GUI / VLM agents, Reporting Protocol=GUI-only2026.06 | 7.6 |