Pricing Plans

Select the plan that suits your project size. Have questions? Contact our team and we'll help you find the right plan.


$1000/month $100/user/month for 1st yr
5 Users (Maximum)
100,000 data items*
10 hours of data logs**
Or 200,000 data items
Email support
Standard Integrations
Talk to a Team Member
Most Popular


10 Users
500,000 data items*
50 hours of data logs**
Or 1,000,000 data items
Slack & Email support
+ 2 Custom Integrations
Talk to a Team Member


Unlimited Users
2,000,000 data items* 200 hours of data logs**
Or 4,000,000 data items
Slack & Email support
+ 5 Custom Integrations
Talk to a Team Member


Contact For Pricing
On-Premise / or VPC Hosted
Unlimited Users
Flexible data items*
Flexible data logs
Dedicated Success Manager Support
Flexible Integrations
Talk to a Team Member
Pricing are for annual contracts. For monthly plans, 20% extra will be charged.

*Data items are images/videos/lidars/etc. that are indexed with annotations, embeddings, and/or metadata.
**$800/month overages charged per extra 20 hours or 200,000 events.

Frequently Asked Questions

Can SceneBox be deployed On-Prem to meet our organizations data requirements?
Yes! SceneBox was built with this flexibility from the beginning. Our Enterprise tier allows you to deploy SceneBox to you VPC or on-prem clusters. SceneBox is a cloud-agnostic micro-services platform built on Kubernetes/Dockers and can be deployed on any supporting infrastructure including AWS, GCS, and Azure.
Do you use customer data, or its derivatives, for your own model training?
Absolutely not. We understand and respect the value of your data and would never compromise its security. As a part of our standard contract, we only use the data that we acquire independently for our internal model training.
On-prem deployment is not a requirement, but how do we know our data is secure?
SceneBox is designed so that your data can stay wherever it resides and SceneBox simply calls the URLs on-the-fly as needed. Whether you use GCP, S3, an internal server, or another storage solution, we’ve got you covered.
How do we integrate SceneBox into our internal data pipelines?
We provide extensive Python clients and Rest APIs. See our documentation for more information.
Is there a free tier for non-commercial usage?
Yes! Under SceneBox’s “scholar program”, individuals can use SceneBox for research and academic purposes free of charge. Please contact us at with a brief description of your application and your academic/scientific affiliation.
Are only images and videos supported? What about other data types?
The best supported data types are images, videos, Lidars, point-clouds, along with any metadata, geo-location, embedding vectors, annotations and time-series data. We also support composite data types such as ROS, RTMaps, KITTI, etc. For a full list of supported data types, click here. As ML engineers ourselves, we recognize the importance of supporting non-standard data (i.e. multi-band images). As such, we have built SceneBox as a data agnostic platform. Novel data types can be accommodated on a case-by-case basis.
How do you provide data search for niche applications?
SceneBox search works with metadata (if available) and the embedding vectors. The embedding vectors are either extracted by a few "vanilla" models that are available on SceneBox platform (e.g., MaskRCNN) or uploaded by the customer. For niche applications where SceneBox's models are not sensitive enough to the features of interest, SceneBox primarily relies on the embeddings that is provided by the user.
Is there a free version that I can test?
My annotator is not listed under your annotations. Can you support new annotators?
Yes we can! We are currently building our arsenal or annotation integrations based on customer requirements. So far we have integrations with CVAT, Scale, LabelBox, Deepen, Dataloop,, V7, SuperAnnotate, Playment, and SageMaker. If you would like an integration, simply let us know and we will complete this as part of your plan.
I only have raw un-annotated data without any metadata. Can SceneBox help?
Yes! This is a common workflow. SceneBox enables you to use embeddings of ML models to organize, index, visualize, and search your raw data. You can either utilize SceneBox's library of models to add embeddings to your data or bring your own embeddings based on your in-house models using our Python Client. We also support multiple embeddings for a datasets.
Do you provide labeling services?
While we do offer the platform and tools required, we do not have a dedicated labeling workforce at this time. There are plenty of great companies out there working on this, many of which we already integrate with! In addition, we host and have tight integration with CVAT, an open-source annotation tool if you are labeling in-house.
Do I need to upload all my raw data to your platform?
You don’t have to! Your data can stay wherever it is. SceneBox's overlay structure allows you to serve data from your cloud buckets on S3, Azure, or GCS. SceneBox also supports multi-cloud data meaning you can access your data centrally while the data is in multiple data lakes.
Is SceneBox only used for real data? How about synthetic? 
SceneBox can be used from data collected from any data source. Either real sensors, or synthetic engines, or mixed. Due to the arbitrary large size of synthetic datasets, data management is even more crucial in synthetic data scenarios.
OK I’m interested. What next?
Great! You can either try the free sandbox version using open source datasets, or you can book a custom demo of SceneBox. If you still have questions, you can contact us.

Bring data to your fingertips with SceneBox today.