Voxel51

What it does

Voxel51 takes in raw multimodal training data, including images, video, 3D point clouds and meshes, sensor streams such as MCAP, medical formats like DICOM and NIfTI, geospatial data, and audio. It indexes that data for natural language, visual similarity, and metadata search, and produces curated datasets, labels, and model evaluation views.

People use it to find where a model fails, inspect those failures in synchronized playback across images, 3D scenes, sensor streams, and annotations, and then slice or filter datasets to remove low-quality samples. Labeling supports automated QA, VLM-assisted labeling, and active learning, and the Voxel51 Agent can run natural-language data tasks and custom plugins.

It suits teams building computer vision or physical AI systems who need to manage large datasets. A free tier offers limited datasets, storage, and inference requests with no credit card required. Team and Enterprise plans cover managed or self-hosted deployment, and Enterprise adds on-premise or air-gapped options. Key limits are the free tier's caps and that paid pricing is not fully published, so Team and Enterprise require contacting sales for exact costs.

What it can do

• Natural language, visual similarity, and metadata search across multimodal datasets
• Synchronized visualization of images, video, 3D scenes, sensor streams, and annotations
• Dataset slicing, querying, and filtering to remove low-quality samples
• Automated annotation QA, VLM-assisted labeling, and active learning
• Model evaluation including prediction vs. ground truth and failure pattern analysis
• Dataset and model versioning with role-based access controls

Before you choose

• Free tier is capped at 3 datasets, 10,000 samples per dataset, 20GB storage, 30,000 inference requests and 30 compute hours per month.
• Self-hosted trials are not self-serve; teams that don't want to upload data to the cloud must contact sales.
• Team plan 'Starting at' figures and Enterprise pricing are not published on the pricing page.

Listed under

• Data
• Computer vision
• Data curation
• Data labeling
• Dataset search
• Model evaluation
• Multimodal data

Get started

Free tier requires no credit card; teams wanting self-hosted trials must contact sales.

1. Sign up for the free tier at app.voxel51.com (no credit card required).
2. Install the open-source FiftyOne library with pip install fiftyone, per the documentation.
3. Import your data or load a zoo dataset, then open the FiftyOne App to explore it.

Free tier has no credit card requirement and monthly usage caps. Team and Enterprise prices are not published; contact sales for a quote.

Source: voxel51.com/pricing, checked 30 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 30 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Voxel51: Find Where Physical AI Models Fail, Fix the Data ↗voxel51.com
• Voxel51 Pricing: Free, Team, and Custom Plans ↗voxel51.com
• FiftyOne Documentation | Datasets and Computer Vision Models ↗docs.voxel51.com
• Voxel51: Find Where Physical AI Models Fail, Fix the Data ↗voxel51.com

Company website

voxel51.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

manual_review

Last checked 30 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.