Graviti
graviti.com
- Free plan
- Web
- API
What Graviti is
Graviti is a data platform for managing unstructured data used in analytics and AI. It brings raw data, metadata, and semantic data together, with customizable filters for querying the latter two. Datasets can be organized with commits and branches, and incremental versioning avoids duplicating data. The platform handles images, video, lidar formats including .pnts, .pcd, .bin, .las, and .ply, as well as time-series and fusion datasets. Users can preview data and annotation distributions online with lists, histograms, and pie charts, without downloading datasets or writing code. Annotation options include Classification, Box, Polygon, Polyline, and Cuboid. Developers can use a Python SDK, OpenAPI, and a command-line interface to create, upload, manage, and read datasets. Workflows can run manually or automatically on commits, and jobs can be written in any programming language. Starter is free; paid plans start at $200/mo, and a 14-day trial is listed.
Who it suits
Graviti is aimed at AI developers who acquire, store, and process unstructured data. Its dataset tools may suit teams working with image, video, lidar, time-series, or fusion data.
Good to know, and what to watch
- Versioned datasets with commits and branches
- Incremental versioning avoids data duplication
- Online previews include lists, histograms, and pie charts
- Python SDK, OpenAPI, and command-line interface
- Role-based dataset access controls
- Paid plans start at $200/mo
- Starter activity logs are kept for 30 days
Plans
Every plan and priceStarter
Standard
Premium
Free plan · 14-day trial
Getting started
- Open the Graviti website
- Create a Starter account or review the 14-day trial
- Use the web interface or API
- Upload and manage datasets with the SDK, OpenAPI or CLI
- Set dataset roles and access approvals
- Create workflows for commit-triggered or manual processing
Limits to check
Starter provides 100GB of private cloud storage, 50K rows and XS /2 hrs free, with activity logs retained for 30 days. Standard and Premium list charges for storage, rows and compute.
The Club's review
Graviti is a platform for organizing and processing unstructured data used in analytics and AI. It is best suited to AI developers and teams working with shared datasets through programmable workflows. Its combination of version control, annotation and workflow tools is the main draw; usage-based charges make the paid plans less predictable than their starting prices suggest.
Overview
Graviti brings raw files, metadata and semantic data into managed datasets, with customizable filters for querying the latter two. That makes it more than a place to store files: teams can organize and inspect data, track changes, and run processing against it within one platform.
Its strongest fit is collaborative work on image, video, lidar, time-series and fusion datasets. For teams that need only general-purpose storage or a database, Graviti’s documented strengths are less relevant.
Key features
Curation and version control
Datasets support commits and branches, with incremental versioning designed to avoid duplicating data. Branches and point-in-time rollback can help teams compare dataset states or return to an earlier one without maintaining separate full copies. Custom filters let users query metadata and semantic data, rather than relying only on filenames.
Data formats and annotations
Graviti supports image and video data, lidar formats including .pnts, .pcd, .bin, .las and .ply, as well as time-series and fusion datasets. Annotation types include Classification, Box, Polygon, Polyline and Cuboid. This range makes it relevant to teams working with varied computer-vision data, though the listed annotation types are specific rather than a general-purpose labeling toolkit.
Preview and workflows
Online previews show data and annotation distributions as lists, histograms and pie charts, without requiring downloads or code. That is useful for quick inspection, while deeper processing can be automated with workflows that run manually or on commits. Workflow jobs can use any programming language and connect through Graviti APIs, which gives developer teams flexibility but assumes they are prepared to build and manage those jobs.
Developer access and controls
A Python SDK, OpenAPI and command-line interface cover dataset creation, upload, management and reading. Graviti also offers GroundTruth Tools for annotation and a marketplace for third-party integrations. Role-based access controls, along with dataset access requests and approvals, help teams govern shared data.
Pricing
Graviti uses a freemium model, with a free plan and a 14-day trial. The Starter plan costs nothing and includes 100GB of private cloud storage, 50K rows and XS /2 hrs free. Its activity logs are retained for 30 days. That is a no-subscription starting point, but the published storage and row allowances make it a limited fit for larger projects.
Standard starts from $200 /mo, with unlimited seats, on-demand storage at $40/TB/mo, $0.1 per 1000 rows/mo and XS at $1/Hour; larger sizes require contacting Sales. Its activity logs are kept for 180 days. It suits teams that need shared access and more capacity, but the base price is only part of the cost because storage, rows and compute are charged separately.
Premium starts from $800 /mo. It includes unlimited seats, on-demand storage at $40/TB/mo, $0.15 per 1000 rows/mo and XS at $1/Hour, with prepaid discounts available by contacting Sales. Activity logs are retained for 1 year. Longer log retention may matter to teams with extended review needs, but the listed per-row rate is higher than Standard’s; teams should compare their expected usage before choosing.
Custom pricing is available, and Graviti invites inquiries about an on-premise plan. The pricing details do not give a renewal term for the trial, so readers should confirm trial terms before relying on it for a project schedule.
Platforms
Graviti is available through web and API access. Those options support both browser-based dataset work and programmatic integration; no desktop or mobile app is part of its stated platform offering.
Who it's for
Graviti is a good match for AI developers and data teams that need versioned, shared unstructured datasets, annotation support, online previews and automated processing. The SDK, API and CLI are particularly useful when teams want to integrate dataset operations into their own code. It is a weaker fit for users who need a simple storage destination or who cannot estimate usage costs across storage, rows and compute.
Pros and cons
Pros
- Dataset history without routine duplication: commits, branches and incremental versioning support controlled changes and rollback.
- Broad data scope: image, video, lidar, time-series and fusion support covers several common AI data types.
- Flexible processing: workflows can run on commits or manually, and jobs can be written in any programming language through Graviti APIs.
- Free entry point: Starter provides private cloud storage and rows at no subscription cost.
Cons
- Usage can add up: paid plans charge separately for storage, rows and compute, so the starting price is not a full cost estimate.
- Starter is capped: 100GB and 50K rows limit how far a substantial dataset project can grow on the free plan.
- Premium raises the row rate: $0.15 per 1000 rows/mo is higher than Standard’s $0.1 rate, despite the higher base price.
- Workflow flexibility requires developer effort: teams must create or integrate processing jobs to take advantage of automation.
Alternatives
Choose Weights & Biases if its model seats, storage and Weave data ingestion allowances better match your needs. Consider Bauplan if you want a free sandbox with CLI, SDK and API access, and can work with public data or make your own uploads public. DagsHub may suit projects organized around repositories, particularly when unlimited public repositories or non-commercial private repositories fit the work. Oxen is another option for repository-based data versioning, with a free Explorer plan that includes public repositories and a limited number of private ones. For teams seeking an open-source data lake and the ability to choose connectors and storage plugins, consider Apache Paimon. ClearML is worth considering if a self-hosted, open-source option is a priority. Choose Databricks Notebooks if a free web workspace with limited serverless compute suits your needs. DataLad is a free, open-source alternative for users on Linux, macOS or Windows. For a broader comparison, see Data Version Control Tools.
Verdict
Graviti is a strong choice for AI developers who need shared, versioned unstructured datasets and want annotation, preview and workflow capabilities alongside them. The free Starter plan offers a practical way to begin, while paid plans add unlimited seats but require careful budgeting for usage charges. Look elsewhere if you need straightforward storage, predictable all-in pricing or a ready-made workflow without developer involvement.
The verdict
Graviti brings dataset management, annotation, preview, and workflow tools together for unstructured data work. A free Starter plan is listed, with paid plans from $200/mo and a 14-day trial.
Questions people ask
Does Graviti have a free plan?
Yes. Starter is listed as free and includes 100GB of private cloud storage, 50K rows and XS /2 hrs free.
What do paid Graviti plans cost?
Standard starts at $200 a month, and Premium starts at $800 a month.
Which platforms can access Graviti?
Graviti provides web and API access.
Can datasets be branched and rolled back?
Yes. Datasets support commits, branches, incremental versioning and point-in-time rollback.
Does Graviti offer a trial?
Yes. The pricing note lists a 14-day trial.
Can Graviti run on-premise?
The deployment model lists both options, and the pricing page invites inquiries about an on-premise plan.
What the maker publishes
- Access controls
- Graviti says datasets can be secured with role-based access control and users can request or approve dataset access. source
- Annotations
- Supported annotation types include Classification, Box, Polygon, Polyline and Cuboid. source
- Data curation
- The platform manages raw data, metadata and semantic data together and supports customizable filters for querying metadata and semantic data. source
- Developer tools
- Graviti provides a Python SDK, OpenAPI and a command-line interface for creating, uploading, managing and reading datasets. source
- Framework integration
- Workflow jobs can be written in any programming language and integrated with Graviti APIs. source
- Integrations
- The platform lists GroundTruth Tools for annotation and says third-party applications can be integrated through its marketplace. source
- Intended users
- The company says its platform is intended to help AI developers acquire, store and process unstructured data. source
- Plan limits
- The pricing comparison lists Starter activity logs kept for 30 days, Standard for 180 days and Premium for 1 year. source
- Product
- Graviti describes its platform as a data platform for managing unstructured data for analytics and AI. source
- Support and custom plans
- The pricing page offers custom plans and invites users to contact Graviti about an on-premise plan. source
- Supported data
- The platform supports image, video and lidar formats including .pnts, .pcd, .bin, .las and .ply, plus time-series and fusion datasets. source
- Version control
- Datasets can be versioned with commits and branches, with incremental versioning that avoids data duplication. source
- Visualization
- Users can preview data and annotation distributions online, including lists, histograms and pie charts, without downloading datasets or writing code. source
- Workflow automation
- Workflows can combine data processing operations and run automatically on commits or through manual operation. source
Checked against the maker's pages on 3 Oct 2026.