OpenCodePapers
computer-use-agents-on-osworld-15-steps
Agents
Computer Use Agents
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
Show papers without code
Paper
Code
Success Rate
↕
ModelName
ReleaseDate
↕
When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents
✓ Link
40.7
Claude Sonnet 4.5 + DeAction
2026-02-09
ARPO:End-to-End Policy Optimization for GUI Agents with Experience Replay
✓ Link
29.9
UI-TARS-1.5-7B + ARPO (extra data)
2025-05-22
Agent S2: A Compositional Generalist-Specialist Framework for Computer Use Agents
✓ Link
27.0
Agent S2 w/ Claude-3.7-Sonnet
2025-04-01
UI-TARS: Pioneering Automated GUI Interaction with Native Agents
✓ Link
22.7
UI-TARS-72B-DPO
2025-01-21
Agent S: An Open Agentic Framework that Uses Computers Like a Human
✓ Link
20.58
Agent S w/ GPT-4o
2024-10-10
ZeroGUI: Automating Online GUI Learning at Zero Human Cost
✓ Link
20.2
ZeroGUI-OSWorld-7B
2025-05-29
UI-TARS: Pioneering Automated GUI Interaction with Native Agents
✓ Link
18.7
UI-TARS-7B-DPO
2025-01-21
Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction
✓ Link
17.04
GPT-4o + AGUVIS-72B
2024-12-05
ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data
✓ Link
16.5
ScaleCUA-32B
2025-09-18
Aria-UI: Visual Grounding for GUI Instructions
✓ Link
15.15
GPT-4o + Aria-UI TH
2024-12-20
Cradle: Empowering Foundation Agents Towards General Computer Control
✓ Link
7.81
Cradle (GPT-4o)
2024-03-05
×