Paperspace Notebooks

Paperspace Notebooks is a hosted Jupyter IDE for data exploration and machine-learning work. It suits developers and teams who want browser-based CPU or GPU compute alongside shared notebooks. Free access makes it easy to start, but the six-hour cap on free machines and paid storage limits matter for sustained projects.

Overview

Notebooks runs Jupyter files in Docker containers, with Python among its supported languages. Users can launch a pre-built template or bring their own, then work on hosted machines from a browser. That is a practical choice when local hardware is insufficient or a team wants a shared environment; it is less attractive if all you need is a notebook editor on your own computer.

Shared persistent storage keeps work available across notebooks and supports collaboration. Teams can invite collaborators, set permissions, fork public projects, and create notebooks. Storage is finite on every listed plan, however, and paid plans charge for usage above the allowance, so large datasets can change the cost of keeping a project in one place.

Key features

Hosted compute and tools

GPU-enabled notebooks run in the browser, and machine choices vary by plan. Pro includes the Free CPU (C4), Free GPU (M4000), P4000, P5000, RTX4000, RTX5000, and A4000 machines. Growth adds A5000, A6000, and A100-80G options. The product supports frameworks and tools including TensorFlow, PyTorch, Keras, Scikit Learn, TensorRT, TensorBoard, ONNX, XGBoost, HuggingFace, RAPIDS, and Fast.ai, giving machine-learning users a broad starting ecosystem.

Storage, data, and collaboration

Pro and Growth subscribers can mount public or private S3-compatible sources, including AWS S3 and DigitalOcean Spaces, directly into notebooks. That integration is useful for teams whose data already lives in object storage, but it is not part of the Free plan. The Paperspace API also allows programmatic management of Paperspace resources through HTTP requests.

Machine auto-shutdown can be set from one hour to one week, while free-tier machines stop at a maximum of six hours. The cap is workable for brief exploration but can interrupt longer sessions; paid plans offer more control over shutdown timing. The Paperspace App is listed for Windows 10+, OS X 10.13+, and Linux beta, with hotkeys, drag-and-drop uploads, and multi-monitor support.

Security and support

Paperspace says its security team monitors threats around the clock and keeps historical activity logs for users and resources. Its security page says datacenters are individually audited against SOC 1, SOC 2, PCI-DSS, and ISO 27001 standards. Standard support is free and ticket-based seven days a week; enterprise support is available by contract and includes infrastructure assistance and customer success managers.

Pricing

The Free plan costs nothing. It includes 5 GB storage, Free CPU (C4) and Free GPU (M4000) machines, and a six-hour maximum auto-shutdown. Free machines are only for Notebooks in private workspaces. This is a useful way to try hosted notebook compute, but the short timeout and smaller storage allowance make it a poor fit for lengthy runs or growing datasets.

Pro (user plan) costs $8 a month, billed $0.0119/hr up to a maximum of $8.00/mo. Pro (team plan) costs $12 a month, billed $0.0179/hr up to a maximum of $12.00/mo. Both include 15 GB storage and the same listed machines, with additional storage at $0.29/GB/mo. The user plan suits an individual who needs paid-plan access; the team plan fits shared work, though its higher cap is worth weighing against the identical listed storage and machine selection.

Growth (user plan) and Growth (team plan) each cost $39 a month, billed $0.0580/hr up to a maximum of $39.00/mo. Each includes 50 GB storage, the Pro machine selection plus A5000, A6000, and A100-80G, and additional storage at $0.29/GB/mo. Growth is the relevant step up for workloads that need those higher-end listed machines or more included storage; for lighter notebook use, Pro has a lower monthly cap. No per-seat allowance is stated for the team plans, so teams should not assume a particular seat count from the plan names.

Platforms

Notebooks is listed for web, API, Linux, macOS, and Windows. Browser access is central to its hosted workflow, while the desktop app adds upload and display conveniences on Windows 10+, OS X 10.13+, and Linux beta. The API is useful for resource management; the product is not limited to a single desktop operating system.

Who it's for

Choose Paperspace Notebooks if you need hosted Jupyter work with GPU options, persistent shared storage, and collaboration in the same environment. It is especially suitable for machine-learning developers whose framework and machine needs fit the listed options, or small teams that want shared notebook projects. Look elsewhere if you require long free machine sessions, need S3-compatible mounts without a paid plan, or want a locally installed notebook without hosted compute.

Pros and cons

• Pros: Free access includes both CPU and GPU machine options, so users can begin without committing to a paid plan.
• Pros: Shared persistent storage, permissions, and project collaboration support work that spans notebooks and teammates.
• Pros: Growth adds A5000, A6000, and A100-80G machines for users whose workloads need those listed options.
• Cons: Free machines shut down after no more than six hours, which limits long-running free sessions.
• Cons: Free storage is 5 GB, and paid plans charge $0.29/GB/mo beyond their included storage.
• Cons: S3-compatible mounts require Pro or Growth, leaving Free users without that integration.

Alternatives

For a fully free and open-source notebook, consider marimo. Deepnote is another freemium option with a free plan that supports up to three editors and five projects. If you prefer a free, open-source notebook installed locally with pip, JupyterLab is a straightforward alternative.

Pluto.jl, Apache Zeppelin, Datalore, SageMath, and Dnote are also options to compare. Browse the broader Notebook Software category to weigh more choices.

Verdict

Paperspace Notebooks is a strong fit for developers and collaborative teams who want hosted Jupyter notebooks with GPU choices, shared storage, and an API in one service. Its free CPU and GPU access is a useful entry point, but a six-hour machine ceiling, 5 GB storage, and paid-only S3 mounts make the free tier better for evaluation than ongoing work. Choose it for managed compute and shared projects; choose a local or free notebook alternative if you do not need hosted machines or want fewer session constraints.