Skip to main content

Credits Usage Guide

Overview

Your usage on the platform is measured in credits, which reflect the amount of AI processing you consume. Credits are primarily calculated based on the number of input tokens and output tokens processed by AI models during a request. In addition to token usage, credits are also consumed by other platform capabilities, including:

  • Knowledge Base operations (e.g., retrieval, vector search)
  • Built-in tools such as:
    • File parsing
    • URL Crawling
    • Text extraction Each of these operations has its own credit cost depending on the processing required.

This document explains how credits are calculated and lists the credit values for all supported models.

What Are Tokens?

Tokens represent pieces of text used by AI models.

  • 1 token ≈ 4 characters
  • ~75 tokens ≈ 1 sentence
  • ~1000 tokens ≈ 750 words

Models charge separately for:

  • Input tokens (your request)
  • Output tokens (AI response)

How Credits Are Calculated

Input Credits  = (input_tokens  / 1000) × input_credit_per_1k
Output Credits = (output_tokens / 1000) × output_credit_per_1k
Total Credits = Input Credits + Output Credits

Credits ensure consistent and predictable billing across models and providers.

KnowledgeBase Credits

The Knowledge Base credit consumption uses a different mechanism. Credits are consumed for:

1. URL-based Knowledge Base

ComponentDescriptionCredits
URL CrawlCost applied per URL crawled5
Embedding CostText extracted from pages is embeddedBased on embedding model credits consumption

2. Document-based Knowledge Base

ComponentDescriptionCredits
Document PageCost applied per page of uploaded document6
Embedding CostText extracted from pages is embeddedBased on embedding model credits consumption

Embedding credits follow the same table as models in the Embedding section below.

Supported Models & Credits

Below are the credit tables grouped by provider within tabs.

ModelInput Credits / 1K TokensOutput Credits / 1K Tokens
chatgpt-4o-latest515
gpt-4o2.510
gpt-4o-mini0.150.6
gpt-51.2510
gpt-5-mini0.252
gpt-5-nano0.050.4
text-embedding-3-small0.130
text-embedding-3-large0.020
text-embedding-ada-0020.100

Summary

  • Credits standardize costs across different AI providers.
  • Each model defines separate rates for input and output tokens.
  • Total credits = input credits + output credits.
  • These tables ensure transparent and predictable usage billing.