Medu
Medu is an AI teacher that grows with each learning project. Learning content is primarily stored on device or in an iCloud workspace, and remote AI features process only necessary data when triggered.Medu 是按学习项目持续成长的 AI 老师。学习内容优先保存在设备或 iCloud 工作区，远程 AI 能力只在你触发时处理必要数据。
Updated: June 8, 2026更新日期：2026年6月8日
Account data: email, password verification, verification codes, optional nickname, authentication token, installation identifier, wallet balance, and session state.账号信息：邮箱、密码验证、验证码、可选昵称、认证 Token、安装标识、钱包余额和会话状态。
Learning data: projects, chat history, Skill Ledger, study cards, cognitive maps, source memory, quizzes, assessment results, handwritten images, and recordings.学习数据：学习项目、聊天记录、Skill Ledger、知识卡片、认知地图、资料记忆、测验、评估结果、手写图片和录音。
DeepSeek is used for teacher replies, study cards, cognitive maps, node expansion, and other text-generation tasks.DeepSeek 用于老师回复、知识卡片、认知地图和节点扩展等文本生成任务。
Qwen / Alibaba Cloud DashScope is used for image answer assessment, speech assessment, and pronunciation/TTS.Qwen / 阿里云 DashScope 用于图片作答评估、语音口语评估和发音/TTS。
The client only receives short-lived AI session tokens from the backend; provider calls, logs, metering, and charging are handled server-side.客户端只向后端领取短期 AI 会话 token；实际模型调用、日志、计量与扣费在服务端完成。
Project packages, media attachments, and cognitive-map ink are written to the Medu workspace; the configured iCloud container is used when available, otherwise local application support storage is used.项目包、媒体附件和认知地图墨迹会写入 Medu 工作区；可用时优先使用配置的 iCloud 容器，否则使用本地应用支持目录。
Images and audio are separated from project JSON into package media files for backup, restore, and migration.图片和音频会从项目 JSON 中拆出为包内媒体文件，以便备份、恢复和迁移。
You can delete learning projects, messages, study cards, cognitive maps, or export/restore workspace data in the app.你可以在应用内删除学习项目、消息、知识卡片、认知地图或导出/恢复工作区数据。
To delete server-side account, wallet, and remote-service records, use Medu's account deletion page and contact support.如需删除服务器侧账号、钱包和远程服务记录，请使用 Medu 的账号删除入口联系支持。
Contact联系
