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

# Plans & rate limits

> Start on Basic with lower limits, unlock higher Basic limits with a credit purchase of at least $5, or request Pro or Enterprise.

New self-serve organizations start on the **Basic** plan, paid per token from [prepaid credits](/docs/credits.md), with lower inference limits until their first qualifying credit purchase. Once a purchase of **at least \$5** settles, your organization automatically gets the higher **Basic** limits below. You can use any supported checkout payment method; saving a credit card or enabling auto-recharge is not required.

You stay on **Basic** with the same token prices. Welcome credits and promotional grants do not qualify. The higher limits stay in place as you spend the credits, but inference still requires an available balance. If your organization joined through a custom invitation or uses custom rate limits, those settings are preserved.

**Pro** offers higher fixed limits through our team. **Enterprise** rate limits are **custom**, configured around your workload. Your current plan appears in the dashboard sidebar on the plan button.

## Inference rate limits

<table>
  <thead>
    <tr>
      <th>Plan</th>
      <th>Requires payment</th>
      <th>Requests / min</th>
      <th>Input tokens / min</th>
      <th>Output tokens / min</th>
      <th>Max concurrent</th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td>Basic</td>
      <td>No</td>
      <td>10</td>
      <td>500,000</td>
      <td>100,000</td>
      <td>3</td>
    </tr>

    <tr>
      <td>Basic</td>
      <td>Yes, at least \$5 in credits</td>
      <td>600</td>
      <td>2,000,000</td>
      <td>500,000</td>
      <td>50</td>
    </tr>

    <tr>
      <td>Pro</td>
      <td>By agreement</td>
      <td>2,000</td>
      <td>5,000,000</td>
      <td>1,500,000</td>
      <td>200</td>
    </tr>

    <tr>
      <td>Enterprise</td>
      <td>By agreement</td>
      <td colSpan="4">**Custom rate limits**, sized per organization to your workload.</td>
    </tr>
  </tbody>
</table>

The limits above apply **per model**: spending your request or token budget on one model never reduces what you can send to another.

## How the limits work

Four dimensions, checked on every request:

* **Requests / min**: how many new requests your organization can start each minute.
* **Input tokens / min**: prompt tokens the models may process per minute.
* **Output tokens / min**: tokens the models may generate per minute, limited separately from input because generation, not reading, is the scarce direction.
* **Max concurrent**: how many requests may be in flight at the same moment.

A request that would push any dimension past its limit is declined with `429`: back off briefly and retry, see [error handling](/docs/inference.md).

## Getting higher limits

To unlock the higher **Basic** limits, open [Billing](https://platform.pearlresearch.ai/billing), click **Add credits**, and complete a purchase of at least **\$5**. Manual top-ups and successful auto-recharges both qualify. Opening checkout, adding a card, or an unsuccessful payment does not change your limits. The updated limits take effect once the payment settles and the network refreshes your organization's settings.

For Pro or Enterprise features and limits, contact our team from the dashboard:

**1. Open the plans dialog**

Sign in and click **Upgrade Plan** in the sidebar (it reads **Your Plan** on organizations already on Pro or Enterprise).

**2. Request a plan upgrade**

Pick **Pro** or **Enterprise** and click **GET A QUOTE**. Describe the plan features and any custom requests, tokens, or concurrency limits you need. Our team reaches out, usually within one business day. Enterprise also unlocks [data residency](/docs/data-residency.md), [monthly invoicing](/docs/credits.md#enterprise), and dedicated compute. The dashboard's **Support** tab reaches the same team. You do not need a quote to unlock the higher Basic limits through a qualifying credit purchase.

> [!TIP]
>
> Bursty traffic, parallel agents, fan-out pipelines? Max concurrent is often the binding limit, and an Enterprise setup can raise it independently of the token limits.


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