Voxel51 Unveils Unified Data Platform 3.0 Aimed at Physical AI Teams

Voxel51 on October 15, 2026 introduced Voxel51 3.0, an end-to-end multimodal data platform for physical AI that brings search, curation, annotation, and evaluation into a single extensible system. Co-founder and CEO Brian Moore, writing in the company’s announcement, called it the company’s “most significant release yet,” and teams can now try the platform on their own data for free at app.voxel51.com.

The release also renames the company’s enterprise product: Voxel51 is the new name for FiftyOne Enterprise, while FiftyOne continues to refer to the open-source project. Moore wrote that the company has spent years working with teams deploying AI in autonomous systems, robotics, manufacturing, and healthcare, and rebuilt the platform over the past several months around the data work that begins once a model moves beyond prototype: investigating failures, curating training sets, and evaluating models across real-world scenarios.

Multimodal Search and Time-Series Support

Voxel51 Search is a persistent hybrid search index that combines semantic retrieval with lexical queries and filters across unstructured content and structured metadata in multimodal datasets. According to the announcement, Search automatically indexes datasets and computes embeddings so queries are available from day one, and it keeps embeddings and structured fields queryable in their native form, combining them at retrieval. For vector retrieval, the platform uses LanceDB, whose columnar architecture supports vector search alongside scalar and full-text indexes while separating storage from compute.

Teams control the embedding pipeline itself, including what data gets embedded, which model and compute to use, and when embeddings are generated, with incremental updates that do not require rebuilding the full dataset. The release adds first-class support for time-series multimodal data: multiple sensor streams can be visualized on the same timeline, and segment-level embeddings let teams retrieve specific moments inside long recordings rather than representing an entire episode with a single embedding.

Voxel51 Agent and Agentic Labeling

Voxel51 Agent chains search, curation, annotation, and evaluation into repeatable workflows built from 45+ MCP tools, 80+ operators, and 18 open-source skills. The announcement describes example uses: a robotics team searching for specific behaviors across thousands of episodes to curate targeted training and evaluation sets; a perception team moving from a model failure to similar longtail scenarios, new annotation, and re-evaluation; and annotation preparation that routes uncertain cases to human reviewers. The Agent can also write code to add new functionality, and all workflows inherit the permissions, security, and governance controls an organization sets.

In a post published on October 8, 2026, ahead of the launch, Voxel51 said the Agent is built on the Model Context Protocol, an open standard for connecting AI applications with external tools, and that it connects to more than 100 model providers, including Anthropic, OpenAI, and Google, with availability across cloud, hybrid, on-premises, air-gapped, and managed deployment options. The company reported that benchmarking the same agent tasks across Claude Sonnet 5, Claude Opus 5, GPT-5.1, and Gemini 3.1 Pro found cost differences of up to 15x.

The 3.0 release additionally unifies annotation workflows for 2D, 3D, video, and multimodal data, with automated QA, annotator metrics, and tools for managing human-in-the-loop review. Its Agentic Labeling capability uses vision-language models to generate labels from natural-language instructions and examples, with review workflows for validating the results. Citing its 2026 State of Physical AI survey, the company said exceptional teams spend more than half their project time on data work at nearly 3x the rate of struggling teams; the survey chart puts those shares at 58 percent of exceptional teams and 21 percent of struggling teams.

Data Provenance Across Four Levels

Voxel51 3.0 adds data provenance at four levels, detailed in a companion post by Chief Architect Ritchie Martori on October 9, 2026. A deployment-level activity log shows who performed an action, when it happened, and what it affected across a deployment; a dataset-level activity log tracks changes within individual datasets; sample-level edit history traces changes to labels, metadata, and other properties; and dataset versioning captures immutable snapshots so teams can compare historical states, reproduce workflows, and identify the exact data used for training or evaluation.

The underlying event architecture uses fiftyone-mq for job scheduling and Activity Core for event tracking, providing a shared foundation for audit logs, annotation metrics, and job monitoring. Permissions and workflow visibility connect each data change to the person or automated job that made it.

Model-Agnostic Design and Nebius Workshops

The platform is model-agnostic by design, so teams can use their own foundation and domain-specific models for embeddings, auto-labeling, inference, and evaluation without being locked into a predefined stack. The announcement states that general-purpose foundation models have learned primarily from internet text and visual data, with limited exposure to the embodied multimodal sensor data that physical systems generate, and that outside a handful of well-funded domains such as autonomous driving, physical AI teams often need to combine foundation models with domain-specific models and their own training data.

Historically, experiencing Voxel51’s enterprise capabilities meant speaking to its team first; with 3.0, self-service access is free at app.voxel51.com, and self-hosted deployments remain available by contacting the company. Voxel51 also said it is collaborating with AI cloud company Nebius, with developers running workflows such as embeddings and auto-labeling on Nebius AI Cloud at hands-on events. The two companies are hosting nine additional Physical AI Workshops across Canada and the United States through November 2026, with registration open.