Skip to content
Varity LogoVarity

Put a cloud inside your product

Give your customers applications, virtual machines and GPUs from your own interface through one deployment API, with privacy controls you can explain to them

Get an API keyTalk to us

Deployments

4 deployments 路 3 live

  • GitHub
    northwind-web

    Dynamic 路 GitHub

    $14.00/moFrankfurt2h ago

    1 vCPU2 GB
  • Docker image
    lumen-api

    Dynamic 路 Docker image

    $28.00/moAmsterdam1d ago

    2 vCPU4 GB
  • harbor-ci

    CPU Virtual Machine

    $0.0192/hrWarsaw3d ago

    2 vCPU2 GB
  • NVIDIA
    atlas-inferenceProvisioning

    GPU Container

    PendingTokyo2m 14s

    NVIDIA1脳 A100

Example: your customers' deployments in the Developer Portal, or in your own product through the API

Priced for platforms,
not per request

Most clouds for platforms

Meter your customers' usage and bill each line separately

  • Bandwidth and egressmetered
  • Requestsmetered
  • Build minutesmetered
  • Seats and add-onsmetered

Varity

A monthly maximum on every app and CPU VM. No separate bandwidth, request or build-minute charges for your customers' applications.

  • Application, per app$7 to $56/mo max
  • CPU VM, per machine$14 to $655/mo max
  • GPU, per machineHourly, live rates

What a platform agreement adds

  • Custom pricing on a committed monthly minimum
  • Higher API rate limits, set in your agreement
Get an API keyTalk to us

Migration

Bring your customers'
workloads with you

Move what your customers already run, one customer at a time, without asking them to change how they build

  1. Bring the image

    Point Varity at the container image you already publish. Private registries work too, with credentials sealed until pull time.

  2. Or bring the repo

    Deploy from a GitHub repository, private ones included, with a token that can read them. The build adapts Vercel project config it finds.

  3. One call per customer

    Create each customer's deployment from your backend. Retry with the same Idempotency-Key and you never get a duplicate.

  4. Test, then switch

    Every deployment gets its own varity.app address. Check it with a real customer task and keep the old host running until you move traffic.

POST /api/deployments
curl -X POST https://varity.app/api/deployments \
  -H "Authorization: Bearer $VARITY_API_KEY" \
  -H "Idempotency-Key: northwind-migrate-1" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "northwind-web",
    "repository": { "url": "https://github.com/your-org/northwind-web" }
  }'
202 Accepted
{
  "status": "deploying",
  "runtime": "dynamic",
  "public_url": "https://northwind-web.varity.app",
  "status_url": "/api/deployments/runs/<run-id>"
}
northwind-webnorthwind-web.varity.appLive

Privacy your customers
can count on

Controls you can name in a security review, not a promise

How Varity handles security
Secrets stay sealed
Environment variables, run commands and Git or registry credentials are encrypted with AES-256-GCM, bound to your account, and deleted when the deployment is torn down
Private when it should be
Keep an application private. It sits behind a password that only your account can read, in the Developer Portal.
No single infrastructure owner
Varity brings independent capacity together as one cloud, so your platform depends less on any one provider

AI Gateway

Offer AI to your customers without holding their prompts

Varity AI Gateway puts the models your customers want behind your product, with privacy you set per request

  1. Your customers' prompts aren't kept

    Prompts, responses and uploads are never stored. Varity keeps only the model, tokens, cost and outcome, for 30 days.

  2. A privacy floor on every request

    Require Private for a customer's request and it only reaches models that meet it. A weaker model is refused, never quietly swapped in.

  3. A key per customer

    Give each customer their own key, with its own privacy permission, allowed models and expiry

  4. One OpenAI-compatible API

    One integration across the live model catalog, instead of one per model provider

AI Gateway is its own product, with its own API key and a prepaid balance kept apart from your Varity Cloud bill

Explore AI Gateway 
POST https://ai.varity.app/v1/chat/completions
{
  "model": "glm-5-3-flash",
  "messages": [ ... ],
  "varity_extensions": { "privacy_floor": "private" }
}

Same floor, three requested models

  • GLM 5.3 FlashPrivateServed
  • DeepSeek V4 FlashPrivateServed
  • GPT OSS 120BAnonymizedRefused
StoredNo prompts, responses or uploadsKept for 30 daysModel, tokens, cost, outcome

Grow your product, not your infrastructure team

Plan migration, customer account mapping, privacy and pricing with us

Get an API keyTalk to usRead the API docs