OpenCodePapers

computer-use-agents-on-screenspot-pro

AgentsComputer Use Agents
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
Click a sortable column header to sort.
PaperCodeAccuracyModelNameReleaseDate
GPT-6 Astra: A new generation of intelligence✓ Link92.7GPT-6 Astra (no tools)2026-09-03
Qwen3.8-Max: A New Bar for Coding and Cowork✓ Link84.5Qwen3.8-Max2026-08-03
GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed✓ Link75.96GroundAnything2026-09-30
UI-Venus-2 Technical Report✓ Link74.1UI-Venus-2-27B2026-08-27
Holo3.1: Fast & Local Computer Use Agents71.5Holo3.1-35B-A3B2026-06-01
ScreenSpot-Pro Leaderboard✓ Link70.4Qwen3.5-122B-A10B2026-02-16
UI-Venus-1.5 Technical Report✓ Link69.6UI-Venus-1.5-30B-A3B2026-02-09
Introducing Gemini 3 Flash69.1Google Gemini 3 Flash (Thinking)2025-12-17
GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives✓ Link65.78GroundingPI2026-09-30
GroundGUI: GUI Grounding by High-Quality Data and Distillation-to-Reinforcement Learning✓ Link63.9GroundGUI-8B2026-10-13
GTA1: GUI Test-time Scaling Agent✓ Link63.6GTA1-32B2025-07-08
EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience✓ Link63.24EvoCUA-OpenCUA-72B2026-01-22
Step-GUI Technical Report✓ Link62.6Step-GUI-8B2025-12-17
Qwen3-VL Technical Report✓ Link62.0Qwen3-VL-235B-A22B-Instruct2025-11-26
ScreenSpot-Pro Leaderboard✓ Link61.9UI-Venus-Ground-72B2025-08-14
UI-TARS: Pioneering Automated GUI Interaction with Native Agents✓ Link61.6UI-TARS-1.5 (early access)2025-01-21
GUI-Libra: Training Native GUI Agents to Reason and Act with Action-aware Supervision and Partially Verifiable RL✓ Link61.4GUI-Libra-4B + Mix Grounding 20K (extra data)2026-02-25
Vero: An Open RL Recipe for General Visual Reasoning✓ Link61.4Vero-Qwen3I-8B2026-04-06
MolmoPoint: Better Pointing for VLMs with Grounding Tokens✓ Link61.1MolmoPoint-GUI-8B2026-03-30
Seed1.5-VL Technical Report✓ Link60.9Seed1.5-VL (thinking)2025-05-11
OpenCUA: Open Foundations for Computer-Use Agents✓ Link60.8OpenCUA-72B2025-08-12
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding✓ Link60.3LocateAnything-3B2026-05-26
POINTS-GUI-G: GUI-Grounding Journey✓ Link59.9POINTS-GUI-G-8B2026-02-06
Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence✓ Link59.3Nemotron 3 Nano Omni2026-04-27
ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data✓ Link59.2ScaleCUA-32B2025-09-18
Mobile-Agent-v3: Fundamental Agents for GUI Automation✓ Link58.0GUI-Owl-32B2025-08-21
VIVAS: Vitalizing Visual Perception in VLM Pre-training via Vision-language Unified Autoregressive Supervision57.1VIVAS2026-09-08
Video2GUI: Synthesizing Large-Scale Interaction Trajectories for Generalized GUI Agent Pretraining✓ Link56.9Mimo-VL-7B + WildGUI2026-05-14
OmegaUse: Building a General-Purpose GUI Agent for Autonomous Task Execution55.47OmegaUse-G2026-01-28
ScreenSpot-Pro Leaderboard✓ Link53.3Qwen2.5-VL-72B-Instruct2025-02-19
Grounding Computer Use Agents on Human Demonstrations✓ Link52.9GroundNext-7B (RL)2025-11-10
LLaDA-UI: Bringing Block-wise Diffusion to Vision-Language GUI Agents✓ Link52.9LLaDA-UI2026-09-09
Kimi-VL Technical Report✓ Link52.8Kimi-VL-A3B-Thinking-25062025-04-10
InfiGUI-G1: Advancing GUI Grounding with Adaptive Exploration Policy Optimization✓ Link51.9InfiGUI-G1-7B (extra data)2025-08-07
STEP3-VL-10B Technical Report✓ Link51.55Step3-VL-10B2026-01-14
FocusUI: Efficient UI Grounding via Position-Preserving Visual Token Selection✓ Link48.3FocusUI-7B (r=100%)2026-01-07
PixelPrune: Pixel-Level Adaptive Visual Token Reduction via Predictive Coding✓ Link47.6Qwen3-VL (2B) with PixelPrune (+KD)2026-04-01
GUI-G$^2$: Gaussian Reward Modeling for GUI Grounding✓ Link47.5GUI-G²-7B2025-07-21
Enhancing Visual Grounding for GUI Agents via Self-Evolutionary Reinforcement Learning✓ Link47.3SE-GUI-7B (extra data)2025-05-18
ScreenSpot-Pro Leaderboard✓ Link45.0UI-AGILE-3B2025-07-29
UI-MOPD: Multi-Platform On-Policy Distillation for Continual GUI Agent Learning✓ Link43.14UI-MOPD2026-07-05
MiMo-VL Technical Report✓ Link41.9MiMo-VL-7B-RL2025-06-04
DeskForge: Dense Supervision from Desktop Environments for Computer-Use Agents✓ Link41.68DeskForge-Qwen3.5-4B2026-10-01
Covering Human Action Space for Computer Use: Data Synthesis and Benchmark✓ Link41.5Phi-Ground-Any-4B + APP Data2026-05-12
Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis✓ Link39.5Jedi-7B2025-05-19
Detect Anything via Next Point Prediction✓ Link36.8Rex-Omni2025-10-14
InfiGUI-R1: Advancing Multimodal GUI Agents from Reactive Actors to Deliberative Reasoners✓ Link35.7InfiGUI-R1-3B (extra data)2025-04-19
BTL-UI: Blink-Think-Link Reasoning Model for GUI Agent✓ Link33.7BTL-UI 7B2025-09-19
UI-R1: Enhancing Efficient Action Prediction of GUI Agents by Reinforcement Learning✓ Link33.5UI-R1-E-3B2025-03-27
UI-Copilot: Advancing Long-Horizon GUI Automation via Tool-Integrated Policy Optimization✓ Link31.6UI-Copilot-7B2026-04-15
AutoWebWorld: Synthesizing Infinite Verifiable Web Environments via Finite State Machines✓ Link27.5AutoWebWorld 7B2026-02-15
ScreenSpot-Pro Leaderboard✓ Link22.9Aguvis-7B-720P2024-12-05
VisionFoundry: Teaching VLMs Visual Perception with Synthetic Images✓ Link20.2Qwen2.5-VL-3B-Instruct + VisionFoundry-10K2026-04-10
ScreenSpot-Pro Leaderboard✓ Link18.9OS-Atlas-Pro-7B2024-10-30
ScreenSpot-Pro Leaderboard✓ Link7.7ShowUI-2B2024-11-26
ScreenSpot-Pro Leaderboard✓ Link3.0MiniCPM-V-2.62024-08-03
ScreenSpot-Pro Leaderboard✓ Link1.0Qwen2-VL-72B-Instruct2024-09-18
ScreenSpot-Pro Leaderboard✓ Link0.1Qwen-VL-7B2023-08-24
Check out how to contribute new results. Then edit this file on GitLab or GitHub.