Qwen-CUA: Native Computer Use for (almost) Everything
modelYour notes
A Qwen × HKU XLANG Lab collaboration (first author Dunjie Lu; Qwen authors include Shuai Bai): a native computer-use agent on a 397B-A17B Qwen MoE backbone that observes only screenshots and acts through keyboard and mouse — no DOM trees, accessibility metadata, or task-specific APIs. The scaffold keeps up to 20 active screenshots and folds older visual history into fixed-size blocks, preserving reusable prompt prefixes for cache efficiency.
Training ran on a cloud rollout fleet with ~100,000 vCPUs and tens of thousands of concurrent environments over ~40,000 verifiable tasks. Scaffold released Apache-2.0 (xlang-ai/Qwen-CUA). Surfaced by the execution-layer sweep as a backfill (posted August 3, missed by the prior cycle); successor territory to XLANG's OpenCUA.