Expired50% confidenceReleaseExact time
XLANG Lab released the OSWorld 2.0 benchmark.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
Valid until: 10/7/2026(expired)
Sources
OSWorld 2.0测评：最强AI Agent完成率仅20%，长流程是致命短板
bilibili熊二等兵7/1/2026
Related Claims
UnverifiedOSWorld 2.0的桌面应用从1.0的LibreOffice、GIMP、VS Code扩展到Slack、LinkedIn、Reaper、WPS、GitLab、Overleaf、Zotero等专业软件64% similarUnverifiedThe LLM 0.32 series is in the alpha stage, and the tool-calling feature is still under active development and testing.55% similarUnverifiedShanghai Artificial Intelligence Laboratory released the Intern-S2 Preview 397B open-source model, making all weights available under the Apache 2.0 license.55% similarUnverifiedThe official documentation for Codex CLI states that it only supports native running on macOS and Linux, but actual testing shows that the current version can be installed and run directly on Windows without WSL.55% similarUnverifiedUiTARS Desktop, a multimodal AI agent desktop application open-sourced by ByteDance, has released version V0.2.0.53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/338167API
curl https://kongchang.com/api/v1/knowledge/claims/338167MCP
get_claim(id=338167)