GUI Agent Task on MemGUI-Bench (Overall)
37.5Precision@1MemGUI-Agent-235B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MemGUI-Agent-235BMemory Type=ConAct, Protocol=zero-shot, Backbone=Qwen3-VL-235B-Thinking2026.06 | 37.5 | 62.5 | 46.8 | |
| M3AMemory Type=Memory Agent2026.06 | 32.8 | 47.7 | 39.3 | |
| Agent-S2Memory Type=Memory Agent2026.06 | 27.3 | 49.2 | 39.5 | |
| Qwen3-VL-235B-ThinkingMemory Type=Action-Thought2026.06 | 24.2 | 46.9 | 30 | |
| Qwen3-VL-235B-InstructMemory Type=Action-Thought2026.06 | 23.4 | 36.7 | 29.2 | |
| MemGUI-8B-SFTMemory Type=ConAct, Protocol=fine-tuned, Backbone=Qwen3-VL-8B-Instruct2026.06 | 23.4 | 35.9 | 30.2 | |
| T3AMemory Type=Memory Agent2026.06 | 22.7 | 42.2 | 29.6 | |
| Qwen3-VL-32B-InstructMemory Type=Action-Thought2026.06 | 20.3 | 32 | 25 | |
| GUI-Owl-1.5-8B-InstructMemory Type=Action-Thought2026.06 | 11.7 | 15.6 | 15.5 | |
| GUI-Owl-1.5-32B-InstructMemory Type=Action-Thought2026.06 | 10.9 | 20.3 | 18.4 | |
| Qwen3-VL-8B-InstructMemory Type=Action-Thought2026.06 | 9.4 | 20.3 | 15.1 | |
| GUI-Owl-7BMemory Type=Action-Thought2026.06 | 6.2 | 10.2 | 5.7 | |
| Mobile-Agent-EMemory Type=Memory Agent2026.06 | 5.5 | 10.2 | 2.4 | |
| UI-Venus-7BMemory Type=Action-Thought2026.06 | 5.5 | 7.8 | 2.6 | |
| Mobile-Agent-V2Memory Type=Memory Agent2026.06 | 3.1 | 3.9 | 0 | |
| AppAgentMemory Type=Action-Thought2026.06 | 3.1 | 9.4 | 1.5 | |
| UI-TARS-1.5-7BMemory Type=Multi-turn2026.06 | 3.1 | 6.2 | 3.8 | |
| SeeActMemory Type=Rule-based2026.06 | 2.3 | 5.5 | 0.2 | |
| CogAgentMemory Type=No History2026.06 | 0 | 0 | 0 |