Pricing and Costs on Varity

Varity Team Core Contributors Updated July 2026

Varity pricing is based on the resources a deployment reserves, not usage metering. For an unchanged deployment profile, your bill does not grow with request volume, bandwidth, or invocation count; changing the profile (resources, services, replicas, accelerators, or app count) can change the price.

Static Sites: First 3 Free

First 3 static sites free, then $1/month per site. Your first 3 active static sites are free for a verified account; each additional active static site is $1.00/month, prorated by the time it is live. Deleting a site frees its slot. Accounts keep a 1 GB of total pinned assets limit. No payment method is required for the free sites, only a verified account. Larger or dynamic apps use Managed Cloud.

Managed Cloud Presets

Dynamic apps select one of four fixed presets. Each price is a monthly maximum: billing is prorated by running time (per minute, against a 730-hour month), so a deployment that runs part of the month bills less than its maximum.

Preset	vCPU	RAM	Ephemeral storage	Price
Starter	0.5	1 GB	2 GB	Up to $7/mo, prorated
Growth	1	2 GB	5 GB	Up to $14/mo, prorated
Scale	2	4 GB	25 GB	Up to $28/mo, prorated
Pro	4	8 GB	50 GB	Up to $56/mo, prorated

Attached services (Postgres, Redis, MongoDB, MySQL, object storage) each bill their own Starter reservation ($7/mo) as a separate reserved deployment, plus the storage you enter for them: $7.00 service + N GB × $0.16 = your monthly maximum per service. Storage defaults to 5 GB and can be set from 1 to 100 GB; Redis is in-memory and takes no volume. Ollama is retired for new deployments.
Persistent storage is an add-on at $0.16/GB-month, prorated on the same period. It bills while allocated, until the volume is deleted.
Each replica reserves its own resources: 2 replicas of a preset bill up to 2x that preset’s maximum.

CPU VMs

CPU VMs bill hourly against a hard monthly cap: the hourly rate is derived from the VM’s monthly maximum divided by 730 hours, and the month’s compute charge never exceeds that cap. Every VM includes a 20 GB NVMe boot disk and one public IPv4, with no separate ingress or egress charge. Additional NVMe storage is $0.09/GB-month, prorated hourly. Every VM carries a two-hour minimum: the first two hours at the quoted hourly rate are booked when the machine is created, and longer runtimes bill by elapsed hour. Billing continues while the VM’s resources remain reserved; stopping the guest OS does not by itself release them; delete the VM to stop billing. VM sizes and regions are published in the Developer Portal as they pass Varity’s end-to-end release verification.

GPU

GPU pricing is a live quote: the exact hourly rate is computed server-side from currently available capacity and shown before you commit. Any continuous-use monthly figure is an estimate for planning, never a cap. GPU machines and GPU containers carry the same two-hour minimum as CPU VMs, booked when the machine or container is created: the minimum for the first 2h is billed at deploy. GPU workloads are paid-only and are excluded from the launch credit.

AI Gateway

The AI Gateway bills the provider’s actual cost for your requests plus a 5% Varity fee, drawn from a prepaid balance. The minimum refill is $20 and the maximum is $1,000; you are charged exactly the balance you are credited, with no funding fee. AI Gateway balances are separate from other Varity products and cannot be spent on Cloud, VM, GPU, or static hosting.

Launch Credit

New accounts with a valid payment method receive a $5 Managed Cloud credit, valid for 14 days. It applies to Managed Cloud CPU deployments only, not to VMs, GPU, the AI Gateway, or static sites (which are already free).

How to Get an Exact Number

Every price is computed by Varity’s server-side pricing authority, so all surfaces show identical numbers:

Developer Portal: developer.store.varity.so shows the exact price before you confirm a deployment.
MCP tool: varity_cost_calculator in your AI editor.