GUI Agent Task on MemGUI-Bench Hard
34.2P@1MemGUI-Agent-235B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MemGUI-Agent-235BMemory Type=ConAct, Protocol=zero-shot, Backbone=Qwen3-VL-235B-Thinking2026.06 | 34.2 | 55.3 | 46.5 | |
| Qwen3-VL-235B-ThinkingMemory Type=Action-Thought2026.06 | 26.3 | 44.7 | 28.2 | |
| M3AMemory Type=Memory Agent2026.06 | 21.1 | 44.7 | 36.4 | |
| Qwen3-VL-235B-InstructMemory Type=Action-Thought2026.06 | 21.1 | 28.9 | 32.2 | |
| MemGUI-8B-SFTMemory Type=ConAct, Protocol=fine-tuned, Backbone=Qwen3-VL-8B-Instruct2026.06 | 21.1 | 34.2 | 32.6 | |
| Agent-S2Memory Type=Memory Agent2026.06 | 18.4 | 36.8 | 36.9 | |
| T3AMemory Type=Memory Agent2026.06 | 18.4 | 34.2 | — | |
| Qwen3-VL-32B-InstructMemory Type=Action-Thought2026.06 | 7.9 | 21.1 | 13.8 | |
| GUI-Owl-1.5-8B-InstructMemory Type=Action-Thought2026.06 | 7.9 | 10.5 | 14.2 | |
| Qwen3-VL-8B-InstructMemory Type=Action-Thought2026.06 | 2.6 | 7.9 | 9.5 | |
| GUI-Owl-7BMemory Type=Action-Thought2026.06 | 2.6 | 2.6 | 2.6 | |
| Mobile-Agent-EMemory Type=Memory Agent2026.06 | 0 | 2.6 | 1.8 | |
| Mobile-Agent-V2Memory Type=Memory Agent2026.06 | 0 | 0 | 0 | |
| SeeActMemory Type=Rule-based2026.06 | 0 | 0 | 0 | |
| AppAgentMemory Type=Action-Thought2026.06 | 0 | 0 | 0 | |
| GUI-Owl-1.5-32B-InstructMemory Type=Action-Thought2026.06 | 0 | 5.3 | 15.3 | |
| UI-TARS-1.5-7BMemory Type=Multi-turn2026.06 | 0 | 0 | 0.9 | |
| UI-Venus-7BMemory Type=Action-Thought2026.06 | 0 | 0 | 0 | |
| CogAgentMemory Type=No History2026.06 | 0 | 0 | 0 |