Run locally Trace every result

OpenScience Your AI research workspace for evidence-backed science

Download OpenScience View research project examples

NEW RESEARCH SESSION

用自然语言启动科研工作

新会话页是第一入口：把问题、数据、文件和项目上下文交给 OpenScience，后续自动沉淀为可追溯 artifact。

OpenScience ✦ Evidence Analysis Review
自然语言任务

分析这批实验数据，生成一份可以复现、可以审查、可以继续迭代的研究报告。

Work in a project 启动
实验数据 dataset.csv
分析代码 analysis.py
研究报告 report.md
本地执行 Artifact 留痕 项目继续推进
LOCAL RESEARCH

Install it where your data lives.

Read files, run analyses, and keep the work on your own computer.

OPEN AND FREE

Free for researchers everywhere.

Use your own models, keep your own files, and start from a desktop workspace.

SCIENCE MODE

Built for many sciences.

350+ research skills cover life science, chemistry, engineering, data science, and causal inference.

REPRODUCIBLE OUTPUT

Results your team can inspect.

Reports, figures, notebooks, and manuscripts keep the evidence and decisions behind them.

EVIDENCE COVERAGE

Start with the evidence base researchers already trust.

Search sources, open references, cite passages, and keep each claim connected to the material that supports it.

11M+ papers searchable, readable, citable, traceable
225K+ regulatory documents searchable, readable, citable, traceable
1M+ clinical trials searchable, readable, citable, traceable
150M+ research abstracts searchable, readable, citable, traceable

EVIDENCE-BACKED SCIENCE

OpenScience turns AI conversations into research artifacts you can reproduce, review, and defend.

Not a single answer. A durable research record: evidence, code, figures, notebooks, manuscripts, and every step that produced them.

SCIENCE MODE

One workspace for all sciences: from molecules to manuscripts, from genomes to causal inference.

10+ research domains and 350+ default skills across life sciences, drug discovery, engineering computation, data science, and social research.

10+ domains 350+ skills reproducible artifacts
Life sciences Genomics · single-cell · multi-omics
Chemistry & drug discovery Molecules · proteins · evidence
Engineering computation Materials · physics · simulations
Data science Statistics · econometrics · causal inference
Policy & social research Surveys · qualitative study · reports
Artifact record Run · trace · reproduce

WORKSPACE MOTION

Evidence, analysis, and artifacts move together.

A research request becomes a living project, not a loose chat transcript.

01

Search evidence, then run the analysis.

OpenScience connects source passages, local files, code execution, and figure output inside one project.

Evidence → Analysis11M+ papers · local code
Query cross-species single-cell RNA-seq integration Sources opened · key passages attached · claim anchors saved
Notebook load matrix → fit model → export figure
02

Keep the project record, not just the answer.

Data, evidence, code, figures, notes, and manuscript drafts stay together so the project can continue later.

PROJECT EXAMPLES

Search evidence. Run analysis. Ship reproducible science.

From evidence to code, from figures to manuscripts, the whole project stays traceable.

Loading evidence report...

CAPABILITY STACK

Research work, organized as artifacts.

Evidence, code, figures, notes, and manuscripts stay in one project record.

01 Evidence

DOWNLOAD

Install where your research happens

Choose your system. The desktop app checks for updates automatically.

macOS

Apple Silicon 与 Intel Mac

安装包准备中

Windows

x64 与 ARM64 安装包

安装包准备中

Linux

DEB / RPM / AppImage

安装包准备中

FAQ

Questions researchers ask first

How do I install and start OpenScience?

Download the package for your system from GitHub Releases, install it, then open OpenScience. macOS uses the DMG, Windows uses the installer, and Debian / Ubuntu can run sudo apt install ./OpenScience-*.deb. If a package is not available yet, run it from source with bun install --frozen-lockfile and bun run start.

Download section Latest package GitHub Releases Install guide Source quick start
Is OpenScience a new model?

No. It is a desktop research workspace around your model, files, tools, and results.

Can I use it on Windows?

Yes. Windows is supported, and the Windows installer will be linked here with the same update flow.

Is it free?

Yes. OpenScience is free and open source. You can connect the model provider you already use.

Does my data leave my computer?

Project files, analysis records, and artifacts stay in your local workspace by default. Content is sent out only when you connect an external model or collaboration service.

Is it only for life sciences?

No. It supports life science, chemistry, drug discovery, engineering computation, data science, social science, policy research, econometrics, and causal inference.

How is it different from Claude Science?

OpenScience is fully open source, works with different model providers, includes medical evidence mode, knowledge deposition, broader research modes, and keeps every result as a reviewable project record.

Can it work with files I already have?

Yes. It can read project files, run analysis, and keep outputs connected to the data and code that produced them.

What does traceable mean?

A result keeps the evidence, code, notes, and changes behind it, so you can reopen it later and check where it came from.

Can it run locally or inside an intranet?

Yes. OpenScience can run on your own computer or private network, with local files, internal data, and private model endpoints.

Will OpenScience train on my ideas or data?

No. OpenScience does not use your project ideas, files, or paper drafts to train models. External model providers follow the terms you choose.

Can my work be interrupted by an account ban?

Your OpenScience projects stay on your machine. If an external model account changes, you can switch to another API, local model, or supported subscription without losing the research record.

Can I use my own API or AI subscription?

Yes. You can connect your own API keys, private endpoints, local models, or supported AI subscription tools, depending on your lab setup.

What can it actually do for a research project?

It can search evidence, read files, run code, make figures and tables, draft reports, prepare manuscripts, and keep each output tied to its sources.

UPDATE FEED

自动更新源

这些 metadata 供客户端自动检查版本；绿色表示该平台已有可用更新源。