CVAT Review, Pricing & Features

CVAT is a free, open-source computer vision annotation tool for images and video. Compare self-hosted vs cloud pricing, features and alternatives.

Category
AI Infrastructure & MLOps
Pricing
Free / Open Source with Freemium SaaS tiers, from Free
Verified
Not yet
Last updated
July 18, 2026
Founded
2018
Headquarters
Palo Alto, California, USA
Free PlanWeb AppAPIOpen SourceAIFreemiumSelf-Hosted

Overview

CVAT began as an internal Intel tool in 2017 and was open-sourced in 2018, later spinning out into independent company CVAT.ai Corporation in 2022. It is one of the most widely adopted open-source platforms for labeling images and video for computer vision and machine learning.

The project offers a free, MIT-licensed, self-hostable Community edition alongside a paid CVAT Online SaaS product and an enterprise self-hosted tier with commercial support.

Key Features

CVAT supports bounding boxes, polygons, polylines, points, cuboids, skeletons, and 3D point-cloud annotation, along with video interpolation and object tracking across frames, and dataset import/export in formats like COCO, YOLO and Pascal VOC.

AI-assisted and automatic annotation features use integrated machine learning models to pre-label data, while role-based access control and a review workflow support quality control across annotation teams.

Pricing

The self-hosted Community edition is free under the MIT License. CVAT Online's free tier includes 1 GB storage and up to three tasks, while the Solo plan starts around $33/month ($23/month billed annually) and the Team plan is priced per user with a two-seat minimum.

An Enterprise self-hosted tier starts at roughly $12,000/year, offering CVAT deployed on a customer's own infrastructure with enhanced security, compliance support and dedicated assistance.

Key Features

Pros & Cons

Pros

  • Free and open source (MIT license) with full self-hosting option
  • Strong video annotation and object-tracking/interpolation tools
  • Wide dataset format support for ML pipelines
  • Active open-source community and OpenCV endorsement

Cons

  • Self-hosting requires Docker/Kubernetes infrastructure expertise
  • Learning curve for teams new to annotation workflows
  • Paid SaaS tiers add up per seat at larger team sizes
  • Enterprise self-hosted tier pricing starts fairly high

Pricing

Frequently Asked Questions

Is CVAT free?

Yes, the self-hosted Community edition is free and open source under the MIT License. CVAT Online also offers a free tier with limited storage and tasks.

Can I self-host CVAT?

Yes, CVAT can be deployed on your own infrastructure via Docker Compose at no licensing cost.

What annotation types does CVAT support?

Bounding boxes, polygons, polylines, points, cuboids, skeletons, and 3D point-cloud annotation for images and video.

Does CVAT support AI-assisted labeling?

Yes, CVAT integrates AI and ML models for semi-automatic and automatic pre-labeling.

Who maintains CVAT?

CVAT.ai Corporation, an independent company spun out from Intel and OpenCV stewardship in 2022, maintains CVAT.

What's the difference between CVAT Community and CVAT Online?

Community is the free, self-hosted open-source edition; CVAT Online is the paid, cloud-hosted SaaS version with managed infrastructure.

Related Tools