A developer encountered an issue when working on a web application in Google AI Studio, where the Gemini API worked without manually entering an API key. However, when saving the project locally and continuing development in Cursor, attempts to use a proxy API address and key (an OpenAI-compatible key supporting Claude, Gemini, and GPT) failed to connect to the Gemini service. Although the proxy key was tested successfully in ChatBox and CherryStudio, it still failed after local replacement. The issue was only resolved after switching to Google’s official API endpoint and key. This experience reveals compatibility issues with AI API proxy services in development tools, reminding developers to pay attention to API key sources and tool differences, providing practical reference value for AI application development.
Original Link:Linux.do

IT资源栈
评论前必须登录!
立即登录 注册