Paperspace Gradient
paperspace.com
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- Browser extension
- API
- Self-hosted
What Paperspace Gradient is
Paperspace Gradient is a machine learning platform for developing, tracking, and collaborating on models. Its browser-based notebook IDE launches GPU-enabled Jupyter notebooks, supports project sharing, and lets users invite collaborators. Notebooks run in portable Docker containers, and GitHub integration supports managing work and compute resources with git. For model serving, deployments create API endpoints with choices for runtimes, instance types, and autoscaling. The platform offers on-demand GPU and IPU instances, with per-second instance pricing described on its product page; utilization of paid instances can add costs beyond plan fees. Gradient says it supports major machine learning frameworks and libraries. Free plans are available, with storage and notebook or instance limits varying by plan. Listed paid options include Pro at $8 a month, plus utilization costs on paid instances. The native Paperspace app is available for Windows 10+, OS X 10.13+, and Linux beta. The platform is described for individuals, researchers, teams, and startups, and includes centralized permissions and activity logs.
Who it suits
Gradient suits individual ML and AI engineers, data scientists, researchers, teams, research groups, and startups that need notebooks, collaboration, or model serving. Its browser-based IDE and desktop app offer different ways to work with projects.
Good to know, and what to watch
- Browser-based IDE launches GPU-enabled Jupyter notebooks.
- Supports project sharing and collaborator invitations.
- Deploys models as API endpoints with autoscaling options.
- Offers free plans with storage and notebook or instance limits.
- Paid instance utilization costs extra beyond plan fees.
- Linux desktop app is listed as beta.
- Free plans have storage and notebook or instance limits.
Plans
Every plan and priceFree (Individual)
Pro (Individual)
Growth (Individual)
Free (Team)
Free plan
Getting started
- Open the Paperspace Gradient website
- Create a free or paid plan account
- Launch a browser-based Jupyter notebook
- Select an available instance type
- Connect GitHub to manage work and compute resources
- Invite collaborators or share a project
- Deploy a model as an API endpoint
Limits to check
The Free plan includes public projects and 5GB storage. T0 includes 10 notebooks, one running notebook, 10GB persistent storage, and low instance types; paid instance utilization costs extra. Paid plans also charge utilization on paid instances.
The Club's review
Paperspace Gradient is a cloud machine-learning workspace combining browser Jupyter notebooks, collaborative projects, model deployment, and GPU or IPU compute. It is best suited to individual ML engineers, data scientists, researchers, and teams that need a route from experiments to API-served models. The subscription tiers are approachable, but storage, notebook quotas, and per-second instance charges make workload sizing essential.
Overview
Gradient brings development and serving into one platform rather than separating notebook work from deployment. Its browser IDE launches GPU-enabled Jupyter notebooks, lets users share projects and invite collaborators, and packages notebooks in lightweight Docker containers for portability. Deployments expose models as API endpoints with configurable runtimes, instance types, and autoscaling. That combination favors projects expected to progress beyond exploration; a user who only needs a local notebook may be paying for capabilities they will not use.
Key features
Notebook development and compute
Containerized notebooks provide a consistent environment while support for major machine-learning frameworks and libraries leaves room for different stacks. GitHub integration connects repositories and compute management to git workflows, a practical advantage for teams already organizing code there. On-demand GPU and IPU instances can match compute to a workload, but they are charged by the second. A low subscription price therefore does not cap the cost of extended training or inference runs.
Collaboration and model serving
Project sharing and collaborator invitations support joint work, while private projects are included in paid Pro and Growth offerings. Serving models as API endpoints, with runtime, instance-type, and autoscaling choices, gives teams a direct production path. The feature set is more substantial than a notebook-only playground, although teams should account for both plan limits and running-instance consumption.
Security, desktop access, and support
Paperspace describes centralized permissions, activity logs, and continuous threat monitoring. Its datacenters meet SOC 1, SOC 2, PCI-DSS, and ISO 27001 standards, which is relevant to organizations with formal compliance requirements. A native app supports Windows 10+, OS X 10.13+, and Linux beta, with hotkeys, drag-and-drop uploads, and multi-monitor support. Free ticket support runs seven days a week; enterprise contracts add infrastructure assistance and customer success managers.
Pricing
Gradient uses a freemium structure, with paid plans starting at $8 per month plus utilization charges on paid instances.
| Plan | Price and allowance | Practical fit |
|---|---|---|
| Free (Individual) | Free; 5GB storage; paid-instance utilization extra | Beginners testing workflows who can work within small storage and usage limits. |
| Pro (Individual) | $8 a month; 15GB storage; paid-instance utilization extra | Individual engineers needing more storage and private work without a large fixed fee. |
| Growth (Individual) | $39 a month; 50GB storage; paid-instance utilization extra | Researchers and practitioners with larger projects and heavier compute needs. |
| Free (Team) | Free; 5GB storage | Teams evaluating collaboration at minimal commitment. |
| Pro (Team) | $12 a month, billed monthly; 15GB storage | Small teams that need a paid collaborative tier. |
| Growth (Team) | $39 a month, billed monthly; 50GB storage | Teams requiring more storage and room for sustained projects. |
| T0 | Free; 10 notebooks, 1 running notebook, 10GB persistent storage, low instance types; utilization extra | Light experimentation with strict concurrency limits. |
| T1 | $12 per user/month; 100 notebooks, 10 running notebooks, 500GB persistent storage, low–mid instance types; utilization extra | Users whose notebook count and storage needs exceed T0. |
| T2 | Custom pricing; low-to-high instance types, private notebooks, scalable storage, unlimited notebooks and running notebooks | Organizations needing uncapped notebook capacity and broad instance selection. |
Another published tier labels the same progression Free, Pro, and Growth: Free provides public projects, 5GB storage, and basic instances; Pro costs $8 a month with private projects, 15GB storage, and mid-range instances; Growth costs $39 a month with private projects, 50GB storage, high-end instances, and expert support. Pro and Growth still add paid-instance utilization costs.
Platforms
Gradient supports API, Linux, macOS, web, and Windows access. The browser IDE is central, while the desktop application extends access across Windows 10+, OS X 10.13+, and Linux beta.
Who it's for
Choose Gradient for individual ML engineers, data scientists, researchers, startups, or research groups that need shared notebooks, version-control integration, and a serving layer in one environment. It is less suitable when fixed, all-inclusive compute pricing or unlimited low-cost storage is the priority.
Pros and cons
Pros
- Notebook-to-API workflow: deployments include runtime, instance, and autoscaling controls.
- Flexible acceleration: on-demand GPU and IPU instances support varied workloads.
- Collaborative organization: sharing, invitations, GitHub integration, and private projects on paid tiers support team processes.
- Compliance posture: SOC 1, SOC 2, PCI-DSS, and ISO 27001 datacenter standards suit regulated environments.
Cons
- Usage billing complicates budgeting because paid instances cost extra per second.
- Entry tiers are constrained by 5GB storage, limited notebook counts, or a single running notebook.
- Private projects and stronger instance options require paid tiers.
Alternatives
Lightning AI Studios is a freemium alternative with a free Studio, 50GB persistent storage, and up to two concurrent GPUs; choose it when those published free allowances fit better.
NVIDIA Triton Inference Server is a free, open-source serving option with development containers; pick it when inference serving, rather than a hosted notebook workspace, is the priority.
ONNX Runtime is free and open source across desktop, mobile, web, and self-hosted platforms, making it preferable when you want a portable runtime rather than a managed Gradient environment.
Amazon SageMaker Autopilot suits teams wanting pay-as-you-go automated model building through API or web access.
Apache TVM is free Apache-licensed software for users seeking an open-source stack.
MATLAB Grader fits MATLAB license holders needing grading workflows.
Deeplearning4j provides fully open-source Apache 2.0 libraries under Eclipse governance.
Ludwig is a free Apache 2.0 open-source option with no paid plans listed.
For broader discovery, browse Deep Learning Software, Data Science Platforms, and Coding playgrounds.
Verdict
Gradient is a strong choice for ML practitioners and teams that want collaborative GPU notebooks and a built-in API deployment path. Choose it for the unified workflow and flexible compute; look elsewhere if per-second instance charges, small free-tier storage, or notebook quotas conflict with your budget or scale.
The verdict
Gradient brings notebook development, collaboration, and API model serving into one machine learning platform. Check the limits of the selected plan and account for paid instance utilization costs when choosing an option.
Questions people ask
Is there a free plan?
Yes. Free plans are listed as free. The Free plan includes public projects and 5GB storage.
What are the paid prices?
Pro is listed at $8 a month and Growth at $39 a month. Paid instance utilization costs are added on plans that specify them.
What platforms are supported?
Gradient supports API, Linux, macOS, web, and Windows. The native app is listed for Windows 10+, OS X 10.13+, and Linux beta.
Can Gradient serve trained models?
Yes. Deployments provide model API endpoints with options for runtimes, instance types, and autoscaling.
Does it support private projects?
Yes. Private projects are listed for paid Pro and Growth plans, while the Free plan lists public projects.
What storage and notebook limits apply?
The Free plan lists 5GB storage. T0 lists 10 notebooks, one running notebook, and 10GB persistent storage; other plans list larger storage or notebook allowances.
What the maker publishes
- Acquisition news
- Y Combinator records DigitalOcean’s acquisition of Paperspace for $111 million in cash on July 6, 2023. source
- API
- Paperspace provides a documented REST API with a unified endpoint schema and improved scalability; legacy Gradient and Core endpoints were deprecated. source
- Auto-shutdown
- Notebook auto-shutdown can be set from one hour to one week, while free-tier machines have a six-hour maximum. source
- Availability claim
- The security page advertises a 99.99% uptime SLA covering network, power, virtual-server, and storage availability. source
- Chrome extension rating
- The Paperspace Chrome extension is rated 0 out of 5 from 0 ratings. source
- CLI licence
- The Gradient CLI is published as open source Python software under the OSI-approved ISC License. source
- Company founding
- G2 lists Paperspace as serving customers since 2014. source
- Company headquarters and size
- Crunchbase lists Paperspace in New York, New York, with 11–50 employees. source
- Company legal name
- Crunchbase lists the legal name as Paperspace, Co. source
- Company ownership
- Paperspace is listed as acquired by DigitalOcean, and the company website now identifies Paperspace as doing business as DigitalOcean. source
- Compliance
- The security page says its datacenters meet SOC 1, SOC 2, PCI-DSS, and ISO 27001 standards. source
- Compute
- The platform offers on-demand GPU and IPU instances, with per-second instance pricing described on its product page. source
- Container integrations
- Gradient can use public or private images from Docker Hub, DigitalOcean Container Registry, and NVIDIA NGC. source
- Containerized notebooks
- Notebooks run in lightweight, portable Docker containers, according to the product page. source
- Deployment options
- full_stack source
- Deployments
- Deployments run container images as hosted machine-learning model endpoints through a REST API. source
- Desktop app
- The native Paperspace app is available for Windows 10+, OS X 10.13+, and Linux beta, with hotkeys, drag-and-drop uploads, and multi-monitor support. source
- Desktop app
- The native Paperspace application supports Windows 10 or newer, macOS 10.13 or newer, and Linux in beta, with hotkeys, drag-and-drop uploads, and multi-monitor support. source
- Documentation depth
- The documentation includes getting-started guides, how-to articles, concepts, reference material, limits, known issues, support pages, CLI references, and SDK/API examples. source
- Frameworks
- Paperspace says Gradient supports major machine learning frameworks and libraries. source
- Free plan
- Yes source
- Funding
- Paperspace announced a $13 million Series A that brought total funding to $19 million. source
- GitHub integration
- Users can connect GitHub to manage work and compute resources with git. source
- Independent evaluation
- The CTO Club rates Paperspace 3.7 out of 5, lists a free plan, and says pricing starts at $8 per month. source
Checked against the maker's pages on 14 Oct 2026.