Agent S4 evolves interfaces and leads open-source OSWorld
With Qwen3.8-27B it scores 21.3% on OSWorld V2, 3.8 times the same model with fixed interfaces.
Stories per day · 30 days
3 RSS
Relatedcomputer-use-agents3gui-agents3open-source3qwen3computer-use2reinforcement-learning2agent-interfaces1agents1agent-s41agent-training1claude1context-management1evolving-interfaces1leash1llm-agents1long-horizon1mai-gym1mobileworld1
24 stories, newest first · one card per story
Stories tagged #osworld
Agent S4 evolves interfaces and leads open-source OSWorld V2
Qwen3.8-27B with evolved interfaces scores 21.3%, above GPT-5.5 and Claude Opus 4.8 at far lower cost.
OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use
OpenGCUA open-sources a generalist computer-use stack that lifts OSWorld-v2 scores and cuts steps.
OS-Helix trains computer-use agents to mix GUI and code
OS-Helix-27B-RL reaches 66.44% on OSWorld-Verified, a further 42.9% relative gain after RL.
LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents
LEASH escalates computer-use agents by latent uncertainty, matching Claude Opus 5 at 19% lower cost.
MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments
MAI-Gym unifies GUI-agent evaluation and online RL; its 9B agent scores 78.4% on MobileWorld.
CarryOn organizes long-horizon agent context around task objects
Authors report gains of 5.8% and 7.0% with token cuts of 51.9% and 22.9% on two benchmarks.
Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction
Work2Skill turns human GUI demos into reusable skills that raise computer-use agent scores.
Claude Haiku 5.5 – Cheapest Model with Enhanced Coding and Computer Use.
Anthropic released Claude Haiku 5.5, about 75 percent cheaper than Haiku 4.5, with much stronger computer use.
SoftwareGym: Scaling Computer-Use Agent Training with Diverse Synthesized Software Environments
SoftwareGym's 182 synthetic apps lift a computer-use agent to 56.97% OSWorld Pass@3.
CUAWright: A Minimal Unified Interface for Digital Agents
CUAWright, a minimal bash-and-filesystem harness, outperforms GUI computer-use agents on several benchmarks.
GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution
GUI-HARVEST evolves GUI agent harnesses, lifting frozen Qwen3-VL by 12.33 points on OSWorld.
DeskForge: Dense Supervision from Desktop Environments for Computer-Use Agents
DeskForge supplies 1.2 million annotated desktop scenes that improve GUI grounding and long-horizon computer-use agents.
ComputerSD: Online Self-Distillation from Real-Time Feedback for Computer-Use Agents
ComputerSD turns real-time GUI feedback into online self-distillation signals for computer-use agents.
OSWorld-Science: A Benchmark of Computer Use Agents for Learning and Using Scientific Software
OSWorld-Science benchmarks 12 VLMs on 146 scientific software tasks with artifact-based scoring.
H Company Releases Holo4: Open-Weight Computer-Use Models That Click, Code and Call Tools Across Desktop, Web, Android and APIs
H Company releases Holo4 open-weight computer-use models (27B dense, 35B-A3B MoE) that operate GUIs, code and tool calls across desktop, web, Android and APIs.
HybridCUA: Learning to Orchestrate GUI and CLI for Computer-Use Agents
HybridCUA-9B combines GUI and CLI actions, reaching 53.6% on OSWorld, up 14.8 points.
OSWorld-Pro: Process-based Evaluation for Computer Use Agents
OSWorld-Pro adds process-level subgoals so computer-use agent failures can be diagnosed beyond final outcomes.
OSWorld-Pro: Process-based Evaluation for Computer Use Agents
OSWorld-Pro scores computer-use agents on over 300 tasks; Claude Opus 5 reaches 75.7%.
MintAct: A Unified Visual Agent for Digital Environments
MintAct vision-language models at 2B, 4B, and 8B unify UI grounding, multi-step navigation, and visual tool use, matching per-domain specialists.
RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments
RSIAgent, a training-free multi-agent framework, builds reusable environment memory enabling Kimi-K3 and GLM-5.3 to beat GPT-6.
UC Berkeley Researchers Release CUA-Lite, an Open Platform Unifying Sandboxes, Data, Evaluation and RL for Computer-Use Agents
UC Berkeley's CUA-Lite is an open platform unifying computer-use agent sandboxes, datasets, evaluation and RL; Lite.OSWorld cuts OSWorld memory 4.1 GB to 0.9 GB.
5
CUA-Universe: A Scalable and Dynamic Environment for Hybrid GUI+CLI Agents
Researchers introduce CUA-Universe, a pipeline turning real desktop software into hybrid GUI+CLI agent environments, lifting a 9B model's OSWorld success rate.
Google releases Gemini 3.8 Flash, its third Flash model in six weeks
Google releases Gemini 3.8 Flash, topping the DeepSWE coding leaderboard six weeks after 3.7 Flash, alongside the cybersecurity-focused 3.8 Flash Cyber.