Cline最新更新:支持使用 Gemini CLI提供的Gemini大模型接口
Cline 更新
🎉 New in v3.18
Optimized for Claude 4: Cline is now optimized to work with the Claude 4 family of models, resulting in improved performance, reliability, and new capabilities.
Gemini CLI Provider: Added a new Gemini CLI provider that allows you to use your local Gemini CLI authentication to access Gemini models for free.
WebFetch Tool: Gemini 2.5 Pro and Claude 4 models now support the WebFetch tool, allowing Cline to retrieve and summarize web content directly in conversations.
Self Knowledge: When using frontier models, Cline is self-aware about his capabilities and featureset.
Improved Diff Editing: Improved diff editing to achieve record lows in diff edit failures for frontier models
注意
• First, you need to install the Gemini CLI tool
• Then, run gemini in your terminal and make sure you Log in with Google
• Only works with personal Google accounts (not Google Workspace accounts)
• Does not use API keys - authentication is handled via OAuth
• Requires the Gemini CLI tool to be installed and authenticated first
Gemini CLI
6 月 25 日,谷歌今日发布博客,宣布 Gemini CLI(命令行界面)现已推出,该工具可让 AI 问答、内容生成等功能直接进入开发者的终端界面。
据谷歌介绍,该工具搭载的是谷歌自家的 Gemini 2.5 Pro 推理模型,支持高达 100 万个 token 的上下文窗口(即 AI 模型可理解的最大信息量);同时与 Gemini Code Assist 代码助手深度集成,内置模型上下文协议(MCP)并支持谷歌搜索。
应用方面,谷歌表示开发人员可利用该工具轻松使用 Gemini,可应用的方面不只编程,在内容生成、任务管理、解决问题等领域皆可应用。
目前,该工具处于预览状态,开发者可通过个人谷歌账户获取免费的 Gemini Code Assist 使用许可证,IT之家附该许可证的使用配额如下:
- 每分钟最多 60 次模型请求。
- 每日最多 1000 次请求。
谷歌并未说明后续是否会收费,也并未说明在超过使用配额后会如何收费。
使用
node 18以上支持
1 | npx https://github.com/google-gemini/gemini-cli |
如果通过web授权失败,可以参考这篇文档,通过配置 GOOGLE_CLOUD_PROJECT 授权
1 | export GOOGLE_CLOUD_PROJECT=gen-lang-client-xxxx |
projectId 可以在Google Cloud Console中查看
Related Issues not found
Please contact @yiGmMk to initialize the comment