Products
Products that prove the research.
Our products are testbeds for our protocols, datasets, and efficient methods — real applications that prove the work at scale.
Applications
Nokuva, Tavoc and Foltrac.
AI-native design editor
The premier AI-native design editor. Describe your vision, refine on a real canvas, build your design system, and convert to production-ready code — all in one place.
- AI generates designs on a real canvas, not code dumps
- The Figma feel — layers, frames, infinite canvas
- Design tokens and live theme resolution built in
Rust-native code platform
Your development environment, supercharged. Tavoc is a high-performance, Rust-native code platform including a complete IDE and local desktop applications, built on our context protocol research.
- Rust-native IDE & desktop apps
- Context-aware assistance
- Efficient by design
Multi-cloud deployment
Deploy anywhere without the complexity. Foltrac handles the infrastructure so you can focus on building — our agent systems research powers its autonomous operations.
- Deploy anywhere
- Autonomous operations
- Infrastructure that manages itself
The platform
Every service, one console.
GPU capacity, inference, training, data, applications and the controls around them. Filter by layer or search, then open any service for its full page.
23 products shown
Compute
GPU capacity on demand or as a dedicated cluster.
- Available
GPU Cloud
NVIDIA H200 by the hour.
$4.79 / GPU-hr - Preview
Dedicated Clusters
Private, interconnected GPU clusters for sustained workloads.
Custom quote - Preview
Sandboxes
Isolated, short-lived environments for developers and AI agents.
Per vCPU, memory and disk hour - Preview
Serverless GPU
GPU containers that scale with requests, down to zero.
Per GPU-second
Inference
Serve open models behind an OpenAI-compatible API.
- Available
Serverless Inference
Open models behind one OpenAI-compatible endpoint.
From $0.10 / 1M tokens - Available
Dedicated Endpoints
Reserved serving capacity for one model, billed per GPU-hour.
$4.79 / GPU-hr - Available
Playground
Send a prompt to any model your workspace can reach.
Billed as inference - Preview
Model Deployments
Immutable deployments with instant promote and rollback.
Billed as inference or per GPU-hour
Training
Fine-tune, train from scratch, and measure what changed.
Data
Managed and synthetic datasets, ready for training.
- Available
Datasets
Managed and synthetic training data, versioned and ready to use.
From $250 / B params / mo - Preview
Data Management
Datasets, connections and storage, with lineage across the platform.
Metered by storage and egress - Preview
Pipelines
Versioned DAGs that automate data, training and deployment.
Billed as the services each step uses - Preview
Data Rooms
Secure document sharing with guests, NDAs and an audit trail.
Included in eligible plans
Applications
Tools built on the platform for designers, developers and operators.
Management & governance
Identity, access, spend and limits for every workspace.
- Available
Identity & Access
Members, roles, policies, SSO and audit for every workspace.
Included - Available
Billing & Cost
Usage, invoices, credits and budgets in one place.
Included - Available
Logs & Monitoring
Request logs, latency and error rates across services.
Included - Preview
Project Canvas
See and change a project's resources on one live canvas.
Included