---
title: AI Usage Limits
description: Rate limits and AI credit limits for Jodit Cloud API keys.
keywords: jodit cloud, ai credits, rate limits, usage limits, startup requests, voice dictation, ai-audio, transcription credits
---

# Usage Limits

The API applies **rate limits** and **AI credit limits** to keep usage fair and the service stable.  
These limits prevent accidental overuse and abuse. Normal applications run without interruption.

There are two main types of limits:

1. **AI Credits**: how much AI processing a key can use. This covers both text generation (`ai-tokens`) and voice dictation (`ai-audio`), each with its own budget.
2. **Startup Requests**: how often the editor loader (`startup`) can be requested.

---

# AI Credits

AI requests are billed using **credits**.

Credits represent the approximate computational cost of AI usage across different models.

In our system:

1 credit ≈ cost of ~1 token on the cheapest AI model

Roughly:

- **1 token ≈ 0.75 English words**
- **1000 tokens ≈ 750 English words**

Because models have different prices, the number of credits consumed per request depends on:

- the model used
- input tokens
- output tokens
- cached tokens

---

## AI Credit Limits

AI usage is limited per time period.

| Period | Limit             | Approximate English Words |
| ------ | ----------------- | ------------------------- |
| Minute | 3,000 credits     | ~2,250 words              |
| Hour   | 30,000 credits    | ~22,500 words             |
| Day    | 300,000 credits   | ~225,000 words            |
| Month  | 3,000,000 credits | ~2,250,000 words          |

These limits are high enough for heavy AI usage but block sudden spikes that could affect system stability.

For example, the **monthly limit** corresponds roughly to processing over **two million English words**.

---

## Audio (Voice Dictation) Limits

Voice dictation in the AI Assistant (the **microphone button** that turns speech into a prompt) is billed separately under the **`ai-audio`** credit budget.

Audio credits are charged from the **speech-to-text (transcription)** cost of your spoken input, independently of the text **`ai-tokens`** budget above. The two budgets do not share limits: heavy dictation never eats into your text-generation allowance, and vice versa.

| Period | Limit           |
| ------ | --------------- |
| Minute | 120 credits     |
| Hour   | 2,000 credits   |
| Day    | 10,000 credits  |
| Month  | 100,000 credits |

The **monthly limit** (~100,000 credits) corresponds roughly to **$100 of speech transcription**, which is many hours of continuous dictation. The per-minute cap smooths out bursts; ordinary back-and-forth dictation never comes close.

`ai-audio` usage appears in its own card in the cloud dashboard, with the same per-minute / hour / day / month breakdown as your other budgets.

---

# Editor Startup Limits

The editor loader (the JavaScript file that initializes the editor) also has request limits.

This protects the service from automated abuse. Normal page loads are not restricted.

| Period | Limit            |
| ------ | ---------------- |
| Minute | 300 requests     |
| Hour   | 10,000 requests  |
| Day    | 50,000 requests  |
| Month  | 500,000 requests |

A single page load usually triggers **one startup request**, so these limits allow tens of thousands of editor loads per day.

---

# Why Limits Exist

These limits keep the service stable and affordable, prevent abuse and automated scraping, and keep usage fair for all users.

Most applications will **never reach these limits during normal usage**.

---

# If You Need Higher Limits

If your application requires higher limits (for example large-scale deployments or high-traffic sites), please contact support and we can adjust the limits for your key.