AI Coding Live News

On October 30, Microsoft's Mage team released Mage-VL-Cowork on Hugging Face under Apache-2.0: a 4B multimodal computer-use agent in which a single model decides at every step whether to act through screenshots and GUI clicks (computer_use) or through bash and Python in a terminal, with no external orchestrator. Microsoft reports a two-run average of 66.0% on OSWorld-Verified (361 tasks, 50-step budget), above Qwen3.6-27B at 63.9% but below the 9B SCALECUA-Qwen3.5-9B at 68.7%.

Zoom

Alibaba Tongyi Labs (Tongyi-MAI) has published and open-sourced OpenGCUA (arXiv:2610.22200), establishing a fully open software stack for Generalist Computer Use Agents. Moving beyond brittle single-click GUI paradigms, OpenGCUA introduces a 'Code-Execution-as-Action' framework combining visual screen perception, accessibility trees, bash scripts, and Playwright APIs into unified programmatic execution blocks. Evaluated across eight frontier models on OSWorld-v2, OpenGCUA boosted mean scores by +19.6 points while cutting task steps by 53.0%, releasing an Apache-2.0 macOS menu-bar harness and Cross-Device SDK.

Zoom

Researchers introduced 'Evolving Agent Interfaces' alongside the Agent S4 meta-framework in arXiv:2610.28601, removing the fundamental bottleneck of computer-use agents (CUAs). Conventional CUAs are forced to operate through human-designed graphical user interfaces—such as clicking spreadsheet cells one by one—inducing excessive action horizons and high failure rates. The new framework allows meta-agents to inspect live applications in scratch environments, dynamically synthesizing, testing, and evolving programmatic API wrappers and tool repositories on the fly. On OSWorld V2 and ScienceBoard, an open-weights 27B model raised task success from 55.03% to 58.58%, rivaling proprietary frontier models at a fraction of the cost.

Zoom

Two new frameworks re-architect how weak and strong models collaborate in long-horizon agents: SpecAgent (arXiv:2610.21463) lets a small model draft multi-step read-only tool trajectories grounded in real observations while holding environment-modifying actions, allowing the strong model to verify the entire segment in a single parallel prefill pass—halving wall-clock time and memory traffic on τ²-Bench. Complementarily, LEASH (arXiv:2610.22280) uses training-free latent-state uncertainty to escalate and supervise computer-use agents, reaching 73.66% on OSWorld with Qwen3.5-35B-A3B—beating always-on Claude Opus 5 (72.90%) while cutting inference cost by 19.2%.

Zoom

Alibaba's Tongyi-MAI team has open-sourced two foundational frameworks for generalist computer-use agents: OpenGCUA (arXiv:2610.22200) and MAI-Gym (arXiv:2610.20939). By unifying GUI interaction with programmatic code execution and harness-to-model decision transfer, OpenGCUA elevates scores on OSWorld-v2 across all eight tested models—including GPT-6 Astra and Claude Opus 5—by +19.6 points while cutting steps by 53.0%. Meanwhile, MAI-Gym scales online RL to 20,000 parallel instances, training MAI-Gym-9B to 78.4% on MobileWorld and 76.2% on OSWorld-Verified.

Zoom

On Oct 7 Anthropic shipped Claude Haiku 5.5 (claude-haiku-5-5), completing the Claude 5.5 family. It targets high-volume, cost-sensitive work such as summaries, compaction, classification and coding subagents, and Anthropic calls it its fastest model at standard speed. Prompts up to 100K cost $0.10 input / $0.50 output per million tokens, and it is the first Haiku with effort levels. It scores 72.4% on the OSWorld 2.1 offline subset (Haiku 4.5: 15.7%) and 39.2% on Terminal-Bench 4.0. Sonnet 5.5 cache reads were cut 50% the same day, and Max/Team plans get monthly API credits.

Zoom

Computer-Use Agents (CUAs) designed to operate graphical desktop environments through virtual keyboard and mouse interactions are traditionally evaluated using outcome-based functional verifiers that assess only the final state after hundreds of operational turns. This black-box approach obscures whether failures stem from syntax typographical errors, GUI bounding-box misalignments, or intermediate state drift, preventing actionable diagnostic insights. NVIDIA Research introduces OSWorld-Pro (arXiv:2609.24890), the first comprehensive process-based evaluation suite for computer-use agents. Spanning 300+ complex desktop workflows decomposed into 2,800+ sequentially dependent subgoals grounded in 67,000+ expert human annotations, OSWorld-Pro deploys calibrated LLM judges to track incremental subgoal completion. Evaluations reveal that OSWorld-Pro presents a formidable challenge to state-of-the-art models: frontier models like Claude Opus 5 achieve only 75.7% (compared to 83.4% on standard OSWorld), systematically isolating critical procedural failure patterns such as exploratory irrelevant actions and click grounding drifts.

Zoom

Existing Computer-Use Agents (CUAs) face a dilemma: relying strictly on graphical user interface (GUI) interactions is slow and error-prone, while building per-application API tools demands unsustainable engineering and lacks generalizability. Researchers from Zhejiang University introduce HybridCUA, combining the UI versatility of GUIs with the execution speed and determinism of the Command Line Interface (CLI). Supported by the newly constructed HybridCUA-8K dataset (5K hybrid trajectories and 3K RLVR tasks), HybridCUA employs a two-stage training paradigm: supervised fine-tuning across interleaved GUI-CLI modalities, followed by CLI-aware reinforcement learning that rewards reliable, selective shell execution. On the OSWorld benchmark, HybridCUA-9B achieves 53.6% accuracy, improving over the base model by 14.8 percentage points, while delivering a 4.0-point gain on WindowsAgentArena.

Zoom

Meta released Muse Glimmer in August, a 30B open-weight local agent model under Apache 2.0 whose quantized weights fit under 20 GB for consumer Macs and single-GPU PCs. On October 19 Meta published its technical report: a ~29.6B dense design distilled from Muse Spark that, by Meta's figures, beats Gemma4-31B and Qwen3.6-27B on tool-use and agent benchmarks such as MCPAtlas, tau3-Banking and SWE-Bench Pro, while still trailing Qwen3.6-27B on SWE-Bench Verified, TerminalBench 2.1 and OSWorld.

Zoom