Unverified50% confidenceFactExact time
早期 GUI Agent 方案依赖计算机视觉截图分析（如 GPT-4V），存在延迟高、定位不精准的问题
1
Sources
50%
Confidence
Long-term
Relevance
7/6/2026
First Seen
Sources
Alibaba Open-Sources Page-Agent: A Web AI Agent for Controlling Interfaces with Natural Language
githubGitHub Trending7/5/2026
Related Claims
UnverifiedIt is estimated that GPT-4's tool-calling capability alone underwent thousands of hours of manual annotation and multiple rounds of iterative optimization.74% similarUnverifiedGPT-4V 等多模态模型在直接输出屏幕坐标时存在显著的空间定位偏差73% similarUnverifiedLarge models such as GPT-4 have demonstrated some vulnerability analysis capability on CTF challenges, but their accuracy in real-world complex codebases remains far below that of human experts.70% similarUnverifiedA developer conducted three rounds of coding capability comparison tests between Cursor + GPT-5 and Buddy + Claude Sonnet 4 using a hospital supply cabinet real-time monitoring system (featuring 3D scene rendering, sensor data monitoring, supply in/out records, and a tech stack of TypeScript frontend and Go backend).70% similarUnverifiedIn programming tests, GPT-4 Thinking can precisely locate Vue front-end bugs, analyze the underlying principles, trace execution flow, and provide complete solutions70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/115341API
curl https://kongchang.com/api/v1/knowledge/claims/115341MCP
get_claim(id=115341)