Open Science 是 Windows、macOS、Linux 上开源的 AI 科研工作台,本地优先、模型可自选。它从一个问题、一篇论文或一份数据文件开始:你用日常语言说清想查什么,AI agent 陪你翻证据、跑分析、出图表。结果怎么来的可以逐段追溯——输入、代码、执行记录都留在工作区里;要分享就把对话分支、文件版本、Notebook 记录与验证证据打包成 .science 文件发出去。适合需要可复现、可审计的科研与数据分析场景。
本次 v1.2.1 更新日志(GitHub 原文):
AIPOCH Open-Science v1.2.1
A unified Resource Center for skills, connectors, and specialists, project-scoped prompt optimization, Codex subscriptions and Claude Fable 5.2 in the built-in Agent, compact evidence for long research sessions, and more reliable Windows updates and cloud-compute sign-in.
AIPOCH Open-Science is an open-source, local-first, model-agnostic AI research workbench for scientists and researchers. It enables reproducible, inspectable research across models with scientific AI agents, Python and R execution, scientific data connectors, and cross-platform support for macOS, Windows, and Linux.
v1.2.1 follows v1.2.0 with a workflow-and-reliability release. Skills, Connectors, Specialists, and their Marketplace move into one unified Resource Center with shared cards, filters, and detail views, and Marketplace installation now works in the Web renderer too. The composer gains manual, project-scoped scientific prompt optimization with review, edit, and undo, plus custom templates in Project settings. The built-in Open-Science Agent can now use Codex subscription accounts for the main agent and subagents, adds Claude Fable 5.2 with a 1M context window, and
…(完整更新日志见项目 GitHub Releases)
