onflip-agent logo

onflip-agent

免费

一款用于编程与日常任务的智能体,由你自己的 ChatGPT 或 DeepSeek 账号提供支持。无需 API 密钥,也不按 token 计费。

访问官网
根据共同标签和关键词推荐grapheaicoollama-workspace-agent
ai-agentagentic-aiaiai-assistantai-coding-assistantai assistantautomation

工具概览

OnFlip 是一款桌面应用,能把聊天账号变成可干活的智能体。它可以读取和编辑你电脑上的文件、运行命令、浏览网页并生成文档,任何有风险的操作都会先征求你的批准。应用驱动你已登录的聊天会话,例如 ChatGPT、DeepSeek、Qwen 或 Google 的 Gemini API,不按 token 计费。一切都在你的机器上进行,没有 OnFlip 服务器、没有遥测,也无需创建账号。应用只通过你自己的浏览器会话与聊天服务通信,不与其他任何对象通信。OnFlip 支持四种服务,可从账号菜单切换,每种服务都有各自的登录、聊天、项目和设置。它提供在项目文件夹中工作、运行命令、浏览网页、同时开两个会话、子任务、Telegram 集成以及撤销/检查等功能。审批模式从只读到不受限制不等,每个副作用都要经过按会话选定的策略。

定价

免费

目录尚未确认完整价格档位,请到官网查看当前额度、限制及计费方式。

访问官网

价格与限制可能变化,请以官网为准,并确认币种、付款周期、最低席位数和用量上限。

应用场景

  • 开发者用 OnFlip 在项目文件夹中工作:它读取代码、编辑代码、运行构建或测试,并报告真实结果。
  • 用户用 OnFlip 完全不指定文件夹地工作,发起聊天来索取文档、电子表格或脚本,生成的文件会出现在对话记录中。
  • 团队用 OnFlip 通过真实的 shell 运行命令,并由自己掌控审批层,从只读到不受限制均可。
  • 用户用 OnFlip 通过真实浏览器浏览网页,可以观看并与智能体操控的页面互动,包括点击、滚动和输入。
  • 用户用 OnFlip 同时运行两个会话,每个窗口都有自己的智能体和浏览器,就像两个聊天标签页。
  • 用户用 OnFlip 从账号菜单在 ChatGPT、DeepSeek、Qwen 或 Gemini API 四种服务之间切换,每种服务都有自己的登录和聊天。
  • 用户用 OnFlip 通过子任务把独立的工作交给第二个智能体,并在子任务面板中查看它做了什么。
  • 用户用 OnFlip 通过 Telegram 在手机上操控同一个智能体,发送任务、批准操作、切换模型和服务。

功能特性

AI 助手

自然语言交互界面

工作流自动化

用户评论

还没有评论,来分享你的使用体验吧

常见问题

Does this break the service's terms?

OnFlip drives a real browser session as a logged-in user and bypasses no checks; Cloudflare challenges are completed by you in a real window. You are responsible for using your accounts within each service's terms.

Why can't it just read my Chrome session?

Chrome and Edge encrypt their cookies with a key bound to the browser itself, and recent Chrome refuses to be driven with its own profile. These are deliberate anti-theft protections, and OnFlip does not go around them. Instead, it signs in inside the real browser on a profile of its own and drives that profile afterwards. Firefox sessions can still be imported directly.

Which model does it use?

On ChatGPT, whatever your account offers, chosen from the model chip: Free and Go accounts default to GPT-5.6 Luna, the plan with unlimited text chats; Plus and Pro run GPT-5.6 Sol in regular chat. On DeepSeek there is one model, with DeepThink on or off. On Qwen, two models that decide for themselves when to think. On Gemini, the chip lists what your own key reports, newest generation first, and defaults to the current Flash. OnFlip sizes its context budget from whichever service and model you are on.

Can I use several accounts?

Yes, one at a time. Switching restarts the app on the service you pick; every sign-in stays valid, so switching back needs no new login.

Does it cost anything per message?

No. Nothing is billed per token — the work costs whatever your account costs, which on ChatGPT's Free and Go plans, on DeepSeek and Qwen, and on Gemini's free API tier is nothing. Limits still apply (image generation and file uploads on ChatGPT; requests per minute and per day on Gemini's free tier), and when a service throttles an account OnFlip waits out the stated pause rather than hammering it.

Do I need a paid plan?

No. A free ChatGPT account works, and free tiers are available on DeepSeek, Qwen, and Gemini API.