What Nextjournal is
Nextjournal is a notebook service for reproducible research and data-driven work, available on the web or self-hosted. A notebook can combine programming-language runtimes, including Python, R, Julia, Clojure, Bash, JavaScript, and ClojureScript, with values exchanged through files. Collaborators can edit notebooks in real time, and automatic versioning saves data and the runtime environment so earlier changes can be restored. Each running notebook gets a compute instance that shuts down when computation is complete. The service supports GPU compute, including up to eight NVIDIA Tesla K80, P100, or P100 Workstation GPUs per runtime. Notebooks can connect to Amazon S3 and Google Cloud Storage, clone private GitHub repositories, and import Docker Hub images. Import formats include Jupyter, IPython, RMarkdown, and Markdown; notebooks can be exported to Markdown. Published notebooks have permanent URLs and can be public or group-limited. The free plan has no hard compute limit, though computation deemed excessive may be cancelled. Paid plans start at $299/month, and a 14-day trial is listed.
Who it suits
It suits machine-learning researchers, data scientists, scientists, professors, and library authors who need notebooks with saved runtime environments or collaborative editing. Self-hosting and enterprise options may also suit organizations with deployment or support requirements.
Good to know, and what to watch
- Combines multiple language runtimes in one notebook.
- Automatic version history includes data and runtime details.
- Real-time notebook collaboration and group publishing.
- Supports GPU compute and cloud-storage integrations.
- Imports several notebook formats and exports to Markdown.
- Free compute is subject to a fair-use policy.
- Team starts at $299/month.
- Enterprise starts at $499/month.
Plans
Every plan and priceFree
Team
Enterprise
Free plan · 14-day trial
Getting started
- Open Nextjournal on the web
- Start with the Free plan or begin the 14-day trial
- Choose hosted use or arrange a self-hosted deployment
- Import a Jupyter, IPython, RMarkdown, or Markdown notebook
- Select the language runtimes and compute resources for the notebook
- Connect storage, repositories, images, or secrets as needed
Limits to check
The Free plan includes unlimited private notebooks but limited concurrent compute resources and entry-level compute under a Fair Use Policy. Team starts at $299 a month with five editors and about 240 USD in computational credits; additional compute is billed per use.
The Club's review
Nextjournal is a hosted notebook workspace for reproducible, data-driven research. It suits research teams, educators and machine-learning practitioners who need several languages, shared editing and preserved environments in one place. The trade-off is a high paid entry price and compute policies that deserve scrutiny for sustained workloads.
Overview
Nextjournal was created by a Berlin-based team that began exploring reproducibility and scientific tooling in 2015. Its hosted execution model provisions a compute instance for each running notebook and shuts that instance down when computation ends, tying resources to active work rather than leaving machines running indefinitely. The result is a research-oriented environment rather than a basic code scratchpad.
The platform is available on the web and as a self-hosted deployment. Its audience includes machine-learning researchers, data scientists, library authors, professors and scientists, with collaboration and publication features that also suit teaching and shared technical work.
Key features
Polyglot notebooks can run Python, R, Julia, Clojure, Bash, JavaScript and ClojureScript runtimes in one document. Values move between runtimes through files, which is useful when a project combines specialist tools or languages, although a single-language workflow may gain little from the extra structure.
Edits, data and the complete runtime environment are saved automatically, and earlier changes can be restored. That preservation is a substantial advantage for reproducibility: collaborators can recover a working state instead of relying on manually maintained code and dependency notes.
Real-time co-editing supports joint work, while public or private groups provide a way to organize access. Published notebooks receive permanent URLs and can be public or restricted to group members. Published versions are read-only, but viewers can remix them, balancing stable reference material with reuse.
Compute options include GPU runtimes supporting up to eight NVIDIA Tesla K80, P100 or P100 Workstation GPUs. Notebooks can access Amazon S3 and Google Cloud Storage, clone private GitHub repositories and import Docker Hub images. Secrets remain outside notebooks and computational environments in Vault and are readable only by the user's profile, a sensible separation for credentials.
Import supports Jupyter, IPython, RMarkdown and Markdown notebooks; export is to Markdown. Those formats make migration practical for existing notebook users, though exported Markdown is not a full round-trip runtime environment.
Pricing
| Plan | Price | What it includes |
|---|---|---|
| Free | Free | Unlimited private notebooks, data and secrets management, limited concurrent compute resources, and entry-level compute under the Fair Use Policy. |
| Team | $299 a month | Five editors, unlimited concurrent compute, high-performance compute, scheduled notebook runs, approximately $240/month in computational credits; additional editors cost $99/month and extra compute is billed per use. |
| Enterprise | $499 a month | Five editors, hosted or on-prem deployment on AWS or Google Cloud, SSO, dedicated compute, priority Slack support, approximately $400/month in computational credits and internal publishing; additional editors cost $125/month. |
A 14-day trial gives prospective customers a short evaluation window. The free plan has no hard compute-time or resource ceiling, but Nextjournal may cancel computation it considers excessive case by case, so it is better for intermittent or moderate work than as an unlimited production queue. Team is the practical threshold for groups needing parallel jobs and scheduling. Enterprise earns its premium for deployment control, identity management and dedicated support; HIPAA guideline compliance may be possible with the on-prem solution using an AWS reference architecture.
Platforms
Nextjournal runs on the web and can be deployed self-hosted. Execution is hosted, with automatic compute lifecycle management and GPU support. Enterprise customers can choose hosted or on-prem installations on AWS or Google Cloud.
Who it's for
Choose Nextjournal for collaborative research where reproducibility, mixed-language notebooks and preserved environments matter as much as writing code. It is a strong fit for laboratory teams, university teaching, data-science groups and machine-learning projects that can use GPU runtimes. Solo users who only need a lightweight notebook, or teams with modest budgets and predictable high-volume compute, should compare lower-cost or self-managed options first.
Pros and cons
Pros
- Reproducibility beyond code: automatic saving includes data and the full runtime environment, with restorable history.
- Broad research workflow: seven language runtimes, cloud storage, private GitHub access and Docker Hub imports cover heterogeneous projects.
- Publication and collaboration: real-time editing, group permissions, permanent URLs and remixable read-only versions support teaching and shared work.
- Serious compute ceiling: GPU runtimes can use up to eight supported NVIDIA GPUs.
Cons
- Expensive paid entry: Team starts at $299 per month before extra editors or usage-based compute.
- Free compute is discretionary: Fair-use enforcement can stop computation judged excessive, making capacity planning uncertain.
- Free concurrency is limited: the no-cost plan is not designed for teams running many notebooks simultaneously.
Alternatives
Deepnote is a better fit when a lower-cost collaborative cloud plan with up to three editors and five projects is sufficient. Datalore suits users who prefer a free cloud allowance measured in CPU hours, storage and parallel notebooks. Hex is worth choosing for data-workspace users wanting broad cell types and a Community plan with small compute. marimo is the straightforward choice for free, permissively licensed open-source notebooks. Paperspace Notebooks is preferable when free CPU and GPU machines are the priority. JupyterHub, JupyterLab and Apache Zeppelin are alternatives for teams that want free, open-source or self-managed notebook software.
For broader browsing, see Notebook Software and Coding playgrounds.
Verdict
Nextjournal is a strong recommendation for research teams that need collaborative, reproducible notebooks spanning several languages and serious compute. Its automatic environment capture, publication controls and GPU support are compelling. Look elsewhere if $299 per month is beyond reach, if you need guaranteed unrestricted free compute, or if an open-source self-hosted notebook already meets your workflow.
The verdict
Nextjournal brings collaborative notebooks, reproducible environments, and multiple runtimes into one research workspace. Check its compute fair-use terms and paid-plan requirements against your workload.
Questions people ask
Is there a free Nextjournal plan?
Yes. The Free plan costs nothing and includes unlimited private notebooks, data and secrets management, and limited concurrent compute resources.
How much do paid plans cost?
Team starts at $299 a month, including five editors. Enterprise starts at $499 a month, also including five editors.
Which languages does Nextjournal support?
Supported languages include Python, R, Julia, Clojure, Bash, JavaScript, and ClojureScript.
Can Nextjournal run on my own infrastructure?
Yes. Nextjournal is available as a hosted service or for self-hosting. Enterprise lists hosted or on-prem deployment on AWS or Google Cloud.
Does Nextjournal offer a trial?
A 14-day trial is listed.
What notebook formats can I use?
Nextjournal can import Jupyter, IPython, RMarkdown, and Markdown notebooks, and export notebooks to Markdown.
What the maker publishes
- Audience
- The homepage describes use cases for machine-learning researchers, data scientists, library authors, professors, and scientists. source
- Collaboration
- Users can edit notebooks together in real time and create public or private groups. source
- Company history
- The About page says the team entered an exploration phase in 2015 and created Nextjournal after exploring reproducibility and scientific tooling. source
- Compute
- The service provisions a compute instance for each running notebook and shuts it down automatically when computation is done. source
- Fair use limit
- The free plan has no hard compute-time or resource limits, but Nextjournal reserves the right to cancel computation it deems excessive case by case. source
- Free plan
- Yes source
- GPU support
- The homepage states support for up to 8 NVIDIA Tesla K80, P100, or P100 Workstation GPUs per runtime. source
- Headquarters
- Berlin, Germany source
- HIPAA
- The pricing FAQ says HIPAA guideline compliance may be possible with the on-prem solution and gives AWS reference architecture as an example. source
- Import and export
- Nextjournal supports importing Jupyter, IPython, RMarkdown, and Markdown notebooks and exporting notebooks to Markdown. source
- Integrations
- Notebooks can access Amazon S3 and Google Cloud Storage, clone private GitHub repositories, and import images from Docker Hub. source
- Polyglot notebooks
- A notebook can use multiple programming-language runtimes, with values exchanged between runtimes using files. source
- Publishing
- Published notebooks have permanent URLs and can be public or limited to group members; published versions are read-only and can be remixed by viewers. source
- Purpose
- Nextjournal describes itself as a notebook for reproducible research and data-driven research. source
- Secrets
- Secrets are stored outside notebooks and computational environments in Vault and can only be read by the user's profile. source
- Paid from
- 299 source
- Support
- Enterprise priority support is provided through a team-to-team Slack channel, with responses within 24 hours during the week. source
- Versioning
- Changes are saved automatically, including data and the full runtime environment, and previous changes can be restored. source
Checked against the maker's pages on 30 Sept 2026.