> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ankra.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Pricing explained

> What Ankra gives you free, what it meters, what a small team pays in a month in three worked examples, where to see your bill and how to cap it.

Ankra charges for the platform by the **worker vCPU-hour**: the CPU of the worker nodes in the clusters it manages, hour by hour, above a free allowance. The servers themselves are billed by whoever runs them. Ankra AI comes with a monthly allowance in every plan. This page explains how those pieces add up. [Billing](https://docs.ankra.ai/platform/billing.md) has the plans, the credit and the invoices.

## What is free

| What | How much | Who gets it |
| - | - | - |
| Worker vCPUs | The first 30 worker vCPUs of the organisation, every hour | Free, Pay as you go, Startup and Enterprise. A custom agreement can set its own |
| Ankra AI | A monthly allowance that grows with the plan. [Usage and the free allowance](https://docs.ankra.ai/platform/ai-provider.md#usage-and-the-free-allowance) has the figures | Each plan |
| The DevOps AI model by Ankra | Its own USD 50 a month, in an organisation Ankra has admitted to the model (a limited set today). There, Ankra AI answers with it once the plan's allowance is used up, unless your own model key or, on Pay as you go, your credit pays instead | Every plan, in an admitted organisation |
| Welcome credit | EUR 59, for 30 days | A Pay as you go organisation, once, when it saves a payment method. Once per card and once per person, and not on top of Ankra Cloud's sign-up credit |
| Pipelines on your own cluster | Not charged | Everyone |
| Playgrounds | Not counted as worker vCPUs | Everyone |

A Pay as you go organisation **without a payment method** is on a smaller free tier. Each month it gets 100 minutes of pipeline steps that Ankra builds on its own build cluster and USD 1 of AI, on Ankra's model key and its own key together. It cannot create or connect clusters, apart from the free trial playground. [Why was I refused?](https://docs.ankra.ai/guides/why-was-i-refused.md#the-free-allowance-without-a-payment-method) has the details.

## What is metered

**Worker vCPU-hours.** Every 15 minutes Ankra reads the CPU of each worker node in the organisation's clusters, both clusters you connected and clusters Ankra built. Control-plane nodes are not counted, and neither are playgrounds or archived clusters. Each hour, the vCPUs above the free 30 are charged at the plan's rate:

| Plan | Per worker vCPU-hour above 30 | With an annual commitment |
| - | - | - |
| Free | Nothing | - |
| Pay as you go | EUR 0.01, taken from your credit | - |
| Startup | EUR 0.01 | 20% less |
| Enterprise | EUR 0.03 | 20% less |

The Enterprise plan is agreed with Ankra and needs a 12-month commitment. [ankra.ai/pricing](https://ankra.ai/pricing) lists the published prices of the Free, Startup and Enterprise plans. Your own plan shows as **Plan** on the billing overview: see [Where to see it](#where-to-see-it).

On the Free plan, until an owner approves paid usage, Ankra refuses to create or grow a cluster it builds for you when the organisation's clusters would then run more than 30 vCPUs. Approving paid usage moves the organisation to Startup. Connecting a cluster is not refused by this limit, but a connected cluster's vCPUs count toward the 30 when a cluster Ankra builds grows. See [the 30-vCPU trial limit](https://docs.ankra.ai/guides/why-was-i-refused.md#the-30-vcpu-trial-limit). Pay as you go has no such limit: worker vCPUs above 30 are charged from your credit.

**Ankra AI above the allowance.** On Pay as you go, once the plan's allowance is spent, Ankra AI keeps answering and the cost is taken from your credit while you have some. On the other plans, and on Pay as you go without credit, AI features on Ankra's model key then pause until the next month. An organisation admitted to the DevOps AI model by Ankra is answered by that model instead, until its own allowance is spent too. AI is counted in US dollars and converted to euros at the day's exchange rate when it is charged. You can also add your own model key under **Ankra AI → Models and provider**.

**Servers.** A cluster Ankra builds on your own cloud account runs on servers your provider bills you for directly. What Ankra Cloud servers cost shows under **Ankra Cloud** on the billing overview.

## Three worked examples

Each example is a small team for one 30-day month (720 hours), on the platform fee only. Server costs are your provider's.

| | 1. A playground | 2. A cluster you connect | 3. A cluster Ankra builds |
| - | - | - | - |
| **What runs** | The free trial playground | 3 workers of 4 vCPUs on your own cloud | 4 workers of 8 vCPUs on your cloud account, plus control-plane nodes |
| **Worker vCPUs** | Not counted | 12 | 32 |
| **Above the free 30** | - | 0 | 2 |
| **Plan** | Any plan, no card | Free, or Pay as you go with a card (Pay as you go needs one to connect) | Pay as you go, or Startup once an owner of a Free organisation approves paid usage (Free cannot build past 30 vCPUs) |
| **Platform fee for the month** | EUR 0 | EUR 0 | 2 vCPUs × 720 hours × EUR 0.01 = **EUR 14.40** |
| **Also** | Nothing to pay | On Pay as you go, saving the card can unlock EUR 59 of welcome credit for 30 days | On Pay as you go the fee comes from your credit. Your provider's bill for the servers |

In all three, Ankra AI within the plan's allowance and pipelines on your own cluster add nothing.

## Where to see it

Open **Settings → Plan and billing** (`/next/organisation/billing`). Its views are:

| View | What it shows |
| - | - |
| **Overview** | Your **Plan**, **vCPU, last 30 days**, **Recent invoices**, **Ankra Cloud** and **Image registry storage**. **Unlock paid usage** shows the prices that apply to your organisation |
| **Usage** | Usage of every meter |
| **Invoices** | Your invoices |
| **Ankra AI** | AI spend against the allowance, and its caps |
| **Credit** | Your credit, top-ups and automatic top-up |

[Usage](https://docs.ankra.ai/platform/usage.md) explains how a usage line reads.

## How to cap it

* **Stay within the free vCPUs.** On the Free plan the platform fee is nothing, and clusters Ankra builds cannot take the organisation past 30 vCPUs until an owner approves paid usage. On Pay as you go, Startup and Enterprise, worker vCPUs above 30 are charged, so keep to 30 or fewer to pay no platform fee.
* **Cap automatic top-ups.** On **Credit**, **Automatic top-up** asks **When the credit falls below (euros)** and **Add this much (euros)**, with **At most this much a day (euros)** and **At most this much a month (euros)**.
* **Cap Ankra AI.** On **Ankra AI**, **Change the caps** sets a **Daily spend cap** (**Your own cap, in US dollars a day**), a **Soft token cap a day** and a **Hard token cap a day**.
* **Count your workers.** The fee follows worker vCPUs hour by hour, so fewer or smaller worker nodes cost less, and a deleted cluster adds nothing.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.