Shared cloud notebook with Python and SQL blocks, an AI agent and an MCP server; the Team plan includes $39 of AI credits per editor each month.
Choose Deepnote if analysts and engineers on your team want to share one notebook, with an agent that keeps a trace of its work. It replaces local Jupyter files passed around by email. Skip it for warehouse-scale compute, where Databricks is larger, or if you need a bigger free AI allowance.
Agents and people work in the same notebook with shared permissions and a trace of every run.
Real-time collaboration plus a workspace where agents and analysts share one set of permissions raises switching cost, while the open-source file format lowers it.
Main trade-off
AI use on the Free plan is capped low, and the compliance controls most companies ask for sit on a custom-priced Enterprise plan.
| Plan | Monthly price | What it includes |
|---|---|---|
| Free | Free | Up to 3 editors, Up to 5 projects, Limited Deepnote AI: 10 completions and 5 Agent calls a month, Basic machines with 5 GB RAM and 2 vCPU, no cap on machine count, 7-day revision history, SOC 2 Type II |
| Team | $39/mo | Unlimited viewers, notebooks and projects, Access to GPT-5.6 and Sonnet 4.6, Premium integrations, Background execution and scheduled notebooks, 30-day revision history, Monthly included AI credits, CPU and GPU compute |
| Enterprise | Custom | Everything in Team, Custom contract and invoice, Priority support and a dedicated success manager, Unlimited AI and bring your own LLM, SSO, directory sync and permission groups, Audit logs, single-tenancy and HIPAA compliance |
Free for up to 3 editors. Team: $39/editor/month on yearly billing, 14-day trial. Enterprise: custom. Monthly billing for Team unchecked.
Deepnote is a collaborative cloud data notebook for Python and SQL, founded in 2019 by Jakub Jurovych in San Francisco. It combines notebooks, dashboards and apps with an in-workspace AI agent and an MCP server, and it open-sourced its core in November 2025. Its pricing page says over 600,000 data pros use it.
Collaborative notebooks
Python, SQL, text and chart blocks that several editors work in at once, with unlimited viewers and notebooks on the Team plan.
Deepnote Agent
An agent that runs data work inside the workspace using a team's skills and rules, with a trace of every run.
Data apps and dashboards
Turns a notebook into an interactive app or dashboard that viewers open without editing rights.
Scheduled notebooks
Runs notebooks on a schedule and in the background on Team and Enterprise.
Open-source core
The .deepnote project format and toolkit are released as open source and run locally in an agentic IDE.
MCP server
Exposes Deepnote notebooks to MCP-capable AI tools through the Deepnote MCP.
Machines and GPUs
Free-plan machines are basic, and Team adds larger CPU and GPU machines with monthly included compute.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| $39/mo | Yes | Web | 3.8 | |
| -- | Yes | Web | 4.2 | |
| $36/mo | Yes | Web | 3.8 | |
| Free | Yes | Web | 4.1 |
Apache-2.0 (open-source core); hosted cloud is proprietary
Deepnote has three plans in US dollars. Free costs nothing for up to three editors. Team is $39 per editor a month billed yearly, with a 14-day trial, and includes $39 of AI credits, $280 of CPU and $50 of GPU a month. Enterprise is custom priced. The monthly-billing figure for Team was not read.
Yes, the Free plan is free forever for up to three editors and five projects. It has unlimited basic machines with 5 GB of RAM and 2 vCPU and seven days of revision history. AI is limited to 10 code completions and 5 Agent calls a month.
An editor is a person who can change notebooks, and Team is priced per editor. Viewers are unlimited on every plan, so a stakeholder who only opens dashboards and apps does not need a paid seat. The free plan allows up to three editors.
Partly. Deepnote open-sourced its core in November 2025, and the GitHub repository carries an Apache-2.0 licence. The cloud workspace with real-time collaboration, the agent and hosting is a paid hosted service.
Deepnote's GitHub description calls it a drop-in replacement for Jupyter with an AI-first design, native data integrations and a cloud mode for real-time collaboration. Jupyter is free and runs locally, while Deepnote adds the shared workspace, scheduling and the agent. The vendor also keeps a head-to-head page with Jupyter.
Deepnote is a collaborative data notebook. A team writes Python and SQL in blocks, charts the results, and shares the notebook as a dashboard or app, with several people editing at once. It was founded in 2019 by Jakub Jurovych, is headquartered in San Francisco, raised a $3.8 million seed round in 2020 and a $20 million Series A in 2022 led by Index Ventures and Accel, per its About page. Its pricing page cites more than 600,000 data professionals. It is filed under data science and ML platforms, and our data science and ML guide places it among the notebook options.
The interesting part in 2026 is the agent. Deepnote Agent works inside the workspace using skills, rules and definitions a team writes down, and every run leaves a trace that anyone with access can open. On the Team plan the pricing page lists access to GPT-5.6 and Sonnet 4.6 (a Claude model), built on work by OpenAI and Anthropic. Deepnote also published an open-source core in November 2025: a readable YAML project format called .deepnote that carries SQL, text and charts beside Python, which you can run locally in VS Code or Cursor.
It fits data teams that want analysts and engineers in the same notebook, and educators whose students collaborate on projects. The pricing page lists SOC 2 Type II on every plan, and HIPAA compliance on Enterprise.
Look elsewhere if you need warehouse-scale compute, where Databricks is the bigger platform, or if you only want a local notebook, where plain Jupyter is free. Hex is the closest hosted rival: its pricing page lists Professional at $36 and Team at $75 per editor, with a focus on published data apps. The free plan is small: three editors, five projects and a handful of AI calls a month.
Product facts on this page (pricing, plans, integrations, compliance) are read from Deepnote's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.