AI Coding Live News

Training long-horizon cowork agents via reinforcement learning requires scalable, iteratively refined interactive environments. Researchers from Fudan University NLP Lab introduced WorkRover, a framework enabling recursive environment evolution. Featuring Factorial Compatibility Replay (FCR) to maintain alignment across work contexts, tasks, and verifiers, and Verifier-to-Turn Credit Assignment (VTCA) to route progress signals to causal prerequisite turns, WorkRover-35B-A3B outperforms baseline checkpoints by 26.27% on JobBench, 29.32% on OfficeQA Pro, and 413.25 Elo on GDPval v2.

Anthropic rolled out a major unified workspace redesign merging Claude Chat and Cowork. Users can now fluidly transition between conversational chat, persistent Cowork pairing, interactive Artifacts, and Design preview modes within a single continuous context thread.

Anthropic is merging Claude Cowork and chat into one Claude experience: ask a quick question or hand off a report, keep work running after you close the laptop, and Claude asks when unclear while you keep final say. Existing Cowork chats, projects, artifacts, connectors, and skills carry over; Claude Design, Slides, and Docs are in beta on paid plans, rolling out to Pro and Max over the next few weeks.

Community coverage notes Microsoft open-sourced Skill Recorder (github.com/microsoft/skill-recorder, ~3.9k stars). After a local screen/activity capture, GitHub Copilot CLI reconstructs an intent plus ordered steps, then builds a SKILL.md Skill or schedulable Automation for Microsoft Scout, Copilot Cowork, and Copilot Studio—preferring native agent tools (gh, web_fetch) over click replay.

Microsoft EVP of Apps & Agents Jeff Teper says OpenAI GPT-6 Astra is rolling out today in Microsoft Copilot Cowork and Copilot Studio, with the official hub post at aka.ms/AstraInCopilot. Teams and custom agents can use the frontier model inside Microsoft’s Copilot stack.

Anthropic enabled Claude to use a user's computer in the background inside Claude Cowork and Claude Code—clicking, typing, and opening apps while the user works on something else. The beta is available to Pro and Max plans in the macOS desktop app. It moves Claude from a chat assistant to a parallel desktop coding and productivity agent.

Command Code says Qwen 3.8 Max 0902 is live: an upgraded Qwen 3.8 Max with ~2.4T parameters, 1M context, and post-training for coding and cowork. It is available on GOAT, Pro, Max, and API—useful when closed-model routes flap.

Claude can now use a computer in the background inside Claude Cowork and Claude Code—clicking, typing, and opening apps while the user works on something else. The beta is available on Pro and Max plans in the macOS desktop app, enabled under Settings → General → Computer use.

Alibaba’s Qwen released Qwen3.8-Max-0902, a 2.4T-parameter, 1M-context model further post-trained for coding and cowork. It debuted at #1 on Code Arena: WebDev with 1,691 points, edging Claude Opus 5, while sitting on the Pareto frontier at a blended per million tokens.

Alibaba Qwen upgraded Qwen3.8-Max to Qwen3.8-Max-0902: about 2.4T parameters, 1M context, and further post-training on Coding and Cowork for complex enterprise, research, and long-horizon workflows. QwenCloud API pricing is $2 input / $6 output per 1M tokens, with explicit/implicit cache hits at $0.17 / $0.25.