Private preview

Request early access.

Running large-model inference? Tell us about your deployment and we will explore whether TokenStorage fits. We are inviting a small number of inference and platform teams into early access to help shape the product.

Work email

Company

Typical model size

GPU nodes and regions

What is the hardest part of distributing models to your inference infrastructure today?

Optional — helps us prioritise

Full name

Role / title

Company website or GitHub org

Primary use case

Primary runtime

Deployment environment

Regions / countries where inference runs Where the bytes actually need to land.

Do you need private models or proprietary fine-tunes?

Design partner program

I'm interested in a design-partner conversation.

Anything else we should know?

We will use this information only to evaluate fit for the preview and contact you about TokenStorage. We do not store your IP address or user agent, and we will not share it with anyone else.

Who we are prioritising

We are not screening people out — smaller teams give us the developer-experience feedback we need. But the first preview cohort is weighted toward teams who feel the problem daily:

Ten or more inference servers, or a credible near-term plan
Models of 50 GB and up
Multi-region, multi-cloud, private, or edge deployments
A real current model source and deployment workflow
Operational pain around scale-out, bandwidth, reliability, or release control

What happens next

We read every submission ourselves.
If there is a fit, we ask for 25 minutes to understand your current workflow and bottlenecks.
Design partners help set the first build priorities — which compatibility surface, which controls, which delivery shape.

We will not promise you a date we cannot meet. TokenStorage is in early development, and the preview opens in cohorts.

Request early access.

You're on the list.