WeaveBench 长程混合接口基准测试，最佳 PassRate 仅 41.2%

View PDF HTML (experimental)

Abstract:Computer-use agents (CUAs) increasingly operate in runtimes that combine visual desktop control, command-line execution, code editing, browsers, and external tools. Existing benchmarks, however, often evaluate these interfaces as separable capabilities, leaving long-horizon cross-interface orchestration under-tested. Thus, we introduce WeaveBench, a long-horizon hybrid-interface benchmark with 114 tasks across 8 real-world work domains, grounded in real user requests and publicly verifiable artifacts. Each task requires agents to combine GUI observations/actions with CLI/code operations within a single trajectory. We evaluate these tasks on a real Ubuntu desktop inside deployed CLI-agent runtimes, augmented with a minimal desktop-control plugin. We also propose a companion trajectory-aware judge that inspects deliverables, files, screenshots, logs, and action traces, while detecting shortcut behaviors such as fabricated visual evidence or hard-coded metrics. Across frontier model-runtime pairings, the best PassRate reaches only 41.2%, showing the benchmark remains far from saturated. The trajectory-aware judge further reveals that outcome-only grading substantially overestimates agent performance. Overall, WeaveBench exposes a critical gap in CUA evaluation and provides an effective testbed to measure whether agents can orchestrate GUI, CLI, and code operations across long-horizon real-world tasks.

Submission history

From: Wanli Li [view email]
[v1] Mon, 8 Jun 2026 12:39:23 UTC (5,458 KB)
[v2] Wed, 10 Jun 2026 16:25:06 UTC (5,467 KB)
[v3] Mon, 6 Jul 2026 14:25:02 UTC (5,459 KB)