When to Compile a Computer-Use Agent? Measuring Payback and Making Compilation Decisions for Token Efficiency

Studies when it pays to compile a GUI procedure that an agent executes repeatedly into a program, given uncertain compilation cost and unknown future reuse. Proposes PACE, which combines a measurement protocol for payback counts with an online compilation algorithm that bounds total cost to at most 1+ε times that of running every task with the agent.

• GUIFlow: Temporal Redundancy-Aware Incremental Inference for GUI Agents
  October 26, 2026 · arXiv
• cua-speedrun: Standardized Benchmarking of the Speed of Computer-Use Agents
  September 30, 2026 · arXiv
• How Many Pixels Is a Digit Worth? Place-Aware Coordinate Entropy for GUI Agent Confidence Estimation
  September 21, 2026 · EMNLP 2026
• TRACE: Trajectory-robust Admission with Evidence Ordering for Efficient GUI Agents
  September 9, 2026 · arXiv
• Efficient GUI Agents: A Systems Survey of Observation, Memory, Action, and Runtime Optimization
  September 2, 2026 · arXiv
• STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models
  June 1, 2026 · arXiv
• Step-level Optimization for Efficient Computer-use Agents
  April 29, 2026 · arXiv
• Tandem: Serving Hybrid Linear-Attention GUI Agents on Mobile NPUs
  October 19, 2026 · arXiv
• STP-GUI: Spatially-Aware Token Pruning for Efficient Web-GUI Agents
  October 14, 2026 · arXiv
• CUAWright: A Minimal Unified Interface for Digital Agents
  October 2, 2026 · arXiv
• Neuro-Symbolic Computer Use: Learning Reusable Policies for Reliable and Efficient Execution
  September 29, 2026 · arXiv
• PhoneCLI: From App Interfaces to Callable Commands for Mobile Agents
  September 28, 2026 · arXiv
• PastForward: Faster On-Device GUI Agents via Computational Experience Reuse
  September 26, 2026 · arXiv
• CUA-Sandbox: Efficient Environments for Computer-Use Agent Reinforcement Learning
  September 26, 2026 · arXiv
• Jev-Mobile: Jev as an Executor for Mobile GUI Agents
  September 24, 2026 · arXiv
• A11y-Compressor: A Framework for Enhancing the Efficiency of GUI Agent Observations through Visual Context Reconstruction and Redundancy Reduction
  May 1, 2026 · arXiv
• Where to Learn and How Far to Roll Out: Efficient Adaptive Training for GUI Agents
  October 29, 2026 · arXiv
• Right Click, Wrong Box: Grading GUI Grounding by What a Click Does
  October 27, 2026 · arXiv
• Learning to Look Beyond Screens: Transferable Visual Observation from GUI Grounding
  October 27, 2026 · arXiv
• From Interaction to Capability: Task Advancement as State and Supervision for GUI Agents
  October 27, 2026 · arXiv
• AdaGround: Token-Normalized Adaptive Zooming with Verifier-in-the-Loop Refinement for GUI Grounding
  October 27, 2026 · arXiv
• PreView: Enhancing GUI Grounding with Preprocessed Views and Cues
  October 24, 2026 · arXiv
• Build or Adopt? Custom-Built vs. General-Purpose LLM Agent Harnesses for Automotive GUI Testing
  October 24, 2026 · arXiv
• IGG: A Benchmark for Interactive GUI Grounding under Visibility Constraints
  October 23, 2026 · arXiv