Subscription plans

A subscription is a fixed recurring charge that grants a workspace a cost-denominated usage window — not a pool of tokens. Every request against a plan window is debited by its actual dollar cost (input/output rate for the model it ran on), so a heavier model drains the window faster than a cheap one and the plan never has to publish a separate token ladder per model. For how usage is measured per capability, see Pricing; this page is about what a plan grants and how it resets.

Two cost ceilings, not one

A plan enforces two budgets at once, both denominated in cost:

• Weekly window — the main budget. It resets on a fixed cadence (a day-of-week and hour set per plan), not a rolling 7 days.
• Five-hour burst guard — a smaller ceiling nested under the weekly one, sized so a short burst of heavy requests can't burn the whole week's budget in one sitting. Unlike the weekly window, it's a rolling look-back — it's always checking the last five hours, not waiting for a fixed reset.

Both scale with the plan's price: a higher tier gets a larger weekly budget and a larger burst guard under it. The Console shows how full each window is as a relative percentage, not an absolute dollar or token figure — the exact window sizes are an implementation detail the API intentionally doesn't expose, so relying on a specific number here would be relying on something that isn't part of the contract.

Per-unit allowances

Beyond the cost windows, each paid tier also bundles a bounded allowance for the count/duration metered capabilities — images generated, embedding tokens, TTS characters, STT seconds, and video seconds (see Pricing for what unit each capability meters). These are separate from the cost windows above: using bundled image or audio allowance doesn't touch your weekly cost budget, and it doesn't carry over — each allowance is bounded by a calendar-fixed monthly and weekly cap, and resets independently of the cost windows. Higher tiers scale every allowance up; Free has none of them. Current per-tier numbers are shown in the Console when you compare plans.

Overflow to prepaid balance

Hitting a plan's weekly ceiling doesn't fail closed by default — it's controlled by an extra usage toggle on the workspace, set in the Console:

• Off (default) — once the weekly window is spent, plan-window requests are rejected until it resets. The five-hour burst guard behaves the same way within its own rolling window.
• On — once the weekly window is spent, further plan requests spill onto the workspace's prepaid balance instead of failing, billed per unit at a cost-plus markup over the underlying rate. This is the same prepaid balance a workspace with no subscription at all runs on entirely.

Per-unit allowances (images, audio, embeddings, video) don't have an overflow toggle the same way — once a bundled allowance for a unit is spent, requests for that unit fall back to prepaid balance automatically, debited at that unit's rate.

Managing a plan

Plans are purchased, upgraded, downgraded, and canceled in the Venna Console, not through the gateway API. A downgrade (including to Free) never takes effect immediately and never refunds the current period — the paid plan stays active until the period ends, then the next renewal switches to the new plan and charges its price. An upgrade takes effect immediately and cancels any pending downgrade.

Next steps

• Pricing — the payment paths and what unit each capability meters.
• x402 — pay per request from a wallet instead of subscribing.
• Quickstart — get a vk_ key and make your first call.