Rerun
Rerun is an open-source SDK and commercial platform designed as a unified data layer for physical AI, particularly robotics and multimodal sensor data.
Our verdict
Who it's for
Data layer for physical AI: visualize, query, transform, and train on multimodal robotics data.
Who should look elsewhere
Commercial Hub requires contacting sales for pricing and setup
Pros
- +Free and open-source SDK with full toolchain for local use
- +Supports multiple programming languages (Python, Rust, C++)
- +Unified data layer covering visualization, query, transform, and training
Cons
- −Commercial Hub requires contacting sales for pricing and setup
- −Primarily designed for physical AI and robotics, may be overkill for simpler data use cases
Key features
- ✓Log, store, and convert multi-rate multimodal data in columnar .rrd files
- ✓SQL and dataframe queries across recordings and catalog
- ✓Declarative visualization framework with interactive viewer (desktop + web)
- ✓PyTorch dataloader for streaming recordings directly to GPUs
- ✓Byte-range indexing for fast lookups across object stores
- ✓Catalog and metadata management with single-tenant isolation
- ✓Auth and SSO for team collaboration
- ✓Deployment in customer's own cloud region with S3-compatible storage
Pricing
| Plan | Price | Billing | Notes |
|---|---|---|---|
| Open source SDK | — | month | Free Apache-2.0 / MIT; Install the SDK pip install rerun-sdk; Python, Rust, and C++ APIs |
| Commercial Hub | — | month | Everything in the SDK; A persistent, managed catalog; Byte-range indexing across your object stores |