Introduction

All plans Nitsor is version control for 3D AI inspection data. It is software for 3D scan labeling, expert review and versioned training datasets for CT and MRI. It keeps model suggestions and people’s decisions with the revisions they changed. The product site is at nitsor.com.

​

What Nitsor is

The product has four parts, with one contract between them:

• A dashboard where a team registers volumes held in its own object storage, defines a taxonomy, applies a pipeline, works a queue, annotates and reviews in a browser editor, and reads each branch’s commit history.
• A control plane that records governed writes as events on a branch, including commits, review verdicts, model Runs and Dataset Versions. It records authorization decisions. Bulk voxels stay in your storage and are read by reference.
• A public HTTP API (/api/v1/) with one OpenAPI description, hashed agent credentials, idempotent writes and asynchronous Requests. It serves records, pipelines, taxonomies, comments, merges, Datasets, Requests and executions. The queue, editor, review verdicts and credential management use the dashboard’s own routes.
• A Python SDK, nitsor, for scripts and training code. Its reader streams a pinned Dataset Version into a training loop and verifies every chunk. Any other language calls the HTTP API.

Architecture

​

The four parts and the one contract between them

Part	What it does
Dashboard	Workspaces, projects, taxonomy, pipeline, queue, editor, viewer, executions, credentials and storage connections. Hosted at app.nitsor.com on Team and Enterprise. The Community self-host stack does not include the dashboard today.
Public API	/api/v1/, described by one OpenAPI 3.1 document.
Python SDK	The Python SDK (nitsor) is not on PyPI yet. Use the HTTP API.
Control plane and storage boundary	Events and references in the control plane; scans and pixel-derived files live in customer or Nitsor-managed object storage.

The dashed line is the data boundary described in Concepts.

​

Plans

Community is free and self-hosted. You run Nitsor on your infrastructure and pay for hosting, storage and compute. Community is available at launch. The self-host package ships at launch. Until then, start on Team. Nitsor’s Python SDK and application code are source-available under FSL-1.1-ALv2, not open source. The bundled Convex backend is licensed separately by Convex under its own FSL-1.1-ALv2. Planned launch pricing for Team is $649 per workspace per month upfront, or $6,228 per year upfront ($519 monthly equivalent), with unlimited people. Both options include 250 model-assisted volumes per monthly usage period, then one flat $0.25 per extra volume. Included volumes reset monthly and do not carry over; yearly Team does not pool them for the year. Extra usage settles after each monthly billing cycle. This billing policy is not active; Plans and billing explains the remaining paid-launch requirements. Enterprise starts at $12,000 a year plus your volume commitment, and adds SSO. Deployment, support, commitments, rollover and overages follow your negotiated contract. The Team rate does not reprice existing Enterprise agreements. Compare plans and billing to choose one.

​

How these pages are arranged

The plan marker at the top of each page names the plans its content applies to and links to the plan comparison.

​

Choose your path

Each path orders the documentation for a different job and links to shared concepts. If a term is unfamiliar, open the Glossary.

​

Next step

Choose a plan in Plans and billing, then follow the Quickstart.

Plans	All plans
Last verified	2026-09-30