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

# Chat Credit System

> How credits are calculated in Geekflare Chat

To make AI usage predictable, [Geekflare Chat](https://geekflare.com/ai/chat/) uses a Credit System instead of confusing token counts.

Every time you send a message or generate an image, a specific number of credits is deducted from your monthly balance based on the model you are using.

## Credit Costs at a Glance

Costs are calculated per request, which includes your prompt, the AI's response, and the context of the conversation.

| Tier                | Cost per Request | Description                                                   |
| :------------------ | :--------------- | :------------------------------------------------------------ |
| **Lite Models**     | **1 Credit**     | Fast, everyday models perfect for drafting and quick answers. |
| **Advanced Models** | **15 Credits**   | Leading models for coding and deep reasoning.                 |
| **Ultra Models**    | **60 Credits**   | The most powerful models available.                           |
| **Standard Images** | **20 Credits**   | High-quality image generation for everyday use.               |
| **HD Images**       | **80 Credits**   | High-definition image generation.                             |
| **Ultra Images**    | **160 Credits**  | Ultra-high-definition premium images.                         |

### Feature Add-ons

If you use advanced workspace features during a chat, a small additional credit fee is applied to that request:

* Knowledge Base (Chatting with files using Data Sources): +5 Credits per request

## Extended Context, Chat History, and File Attachments

The standard credit costs listed above cover a base limit of up to **10,000 tokens** per request (roughly 40,000 characters, or 7,500 to 10,000 words).

To give you high-quality answers, the AI needs context. Your total token count for a single request includes:

1. Your current message
2. Previous messages in the same conversation (Chat History)
3. Any files, documents, or images you attach directly to the chat
4. The AI's response

**The Multiplier Effect**
If your conversation gets very long or you attach large files, your request may exceed the 10,000 base token limit. When this happens, a multiplier is applied to the base cost of your chosen model.

We calculate the multiplier using this formula:

Multiplier = ⌈ Total Tokens / 10000 ⌉

**How File Attachments Work**

* **Text & Data Files** (`.txt`, `.csv`, `.json`, `.py`, etc.): Tokens are calculated based on the text length.
* **PDFs & Images:** Token usage is dynamic. Multi-page PDFs and high-resolution images require more tokens to process, which may trigger a multiplier.

<Note>Tip: If you need to analyze a document without triggering high multipliers, add it to your Knowledge Base (RAG) in Data instead of attaching it directly. The Knowledge Base searches only the most relevant sections, keeping your token count low.</Note>

## View Models by Tier

<AccordionGroup>
  <Accordion title="Lite Models (1 Credit)">
    * **Google:** Gemini 3.1 Flash Lite
    * **OpenAI:** GPT 5 Nano, GPT 5 Mini, GPT 5.1 Codex Mini, GPT 4.1 Nano, GPT 4.1 Mini, GPT 4o Mini
    * **Mistral:** Mistral 14B, Mistral Magistral Small
    * **MiniMax:** Minimax M2.7, Minimax 2.5
    * **Qwen:** Qwen 3.5 Flash
    * **Meta:** Llama 4 Scout, Llama 4 Maverick
    * **DeepSeek:** DeepSeek V4 Flash
  </Accordion>

  <Accordion title="Advanced Models (15 Credits)">
    * **Anthropic:** Claude Haiku 4.5, Claude Sonnet 4.6
    * **Google:** Gemini 3.5 Flash, Gemini 3.1 Pro, Gemini 3 Pro, Gemini 3 Flash
    * **OpenAI:** GPT 5.4, GPT 5.3 Chat, GPT 5.3 Codex, GPT 4.1, o4 Mini, o3, GPT 4o
    * **xAI:** Grok Build 0.1, Grok 4.3, Grok 4.20
    * **Perplexity:** Sonar, Sonar Reasoning
    * **Mistral:** Mistral Large, Mistral Magistral Medium
    * **Qwen:** Qwen 3.7 Plus, Qwen 3.7 Max, Qwen 3.6 Max, Qwen 3.6 Plus
    * **Kimi:** Kimi K2.6, Kimi K2.5
    * **DeepSeek:** DeepSeek V4 Pro
    * **MiniMax:** MiniMax M3
  </Accordion>

  <Accordion title="Ultra Models (60 Credits)">
    * **Anthropic:** Claude Opus 4.7, Claude Opus 4.6
    * **OpenAI:** GPT 5.5
    * **Perplexity:** Sonar Pro
  </Accordion>

  <Accordion title="Standard Images (20 Credits)">
    * **xAI:** Grok Imagine Image
  </Accordion>

  <Accordion title="HD Images (80 Credits)">
    * **Google:** Gemini 3.1 Flash Image (Nano Banana 2)
    * **OpenAI:** GPT Image 2
    * **xAI:** Grok Imagine Image Pro
  </Accordion>

  <Accordion title="Ultra Images (160 Credits)">
    * **Google:** Gemini 3 Pro Image (Nano Banana Pro)
    * **OpenAI:** GPT Image 1
  </Accordion>
</AccordionGroup>

## How Credits are Calculated (Examples)

Here is how your credits are used in real-world scenarios:

* **Scenario A (Standard Chat)**: Asking a Lite Model like GPT 5 Nano to write an email. (Under 10K tokens).
  * Cost: **1 Credit**.
* **Scenario B (Advanced Task)**: Asking an Advanced Model like Claude Sonnet 4.6 to write a React component. (Under 10K tokens).
  * Cost: **15 Credits**.
* **Scenario C (Long Chat History)**: You have been chatting with an Advanced Model for an hour. The accumulated chat history sent to the AI is now 25,000 tokens. The formula applies a 3x multiplier to the base 15-credit cost.
  * Cost: **45 Credits**.
* **Scenario D (Large File Attachment)**: You attach a large 40,000-token file to a Lite Model (1 credit). The formula applies a 4x multiplier.
  * Cost: **4 Credits**.
* **Scenario E (Knowledge Base)**: Asking a Lite Model to summarize a small PDF you uploaded to your Knowledge Base.
  * Cost: 1 (Lite Model) + 5 (Knowledge Base Add-on) = **6 Credits**.

## Tips to Reduce Credit Usage in Geekflare Chat

If you want to stretch your monthly credits further, follow these best practices.

**Start new chats frequently**

Because the AI needs to remember what you discussed, every new message you send includes your previous chat history. In a long conversation, this accumulating history can push your total tokens past the 10,000 base limit, triggering a 2x, 3x, or higher multiplier. When you change topics or no longer need the AI to remember earlier messages, start a fresh chat.

**Use Lite models for everyday tasks**

Not every prompt requires the heavy reasoning of a premium model. Use Lite models for drafting emails, brainstorming, summarizing text, or answering general questions. Reserve Advanced and Ultra models for coding, deep analysis, and intricate logic.

**Upload recurring files to your Knowledge Base**

If you attach a large file directly in the chat window, its entire token weight is processed with every single message you send. If you plan to chat with a file multiple times, upload it to your Knowledge Base (Data) instead.

**Combine multiple instructions into one prompt**

Since credits are charged per request, sending three separate, short messages will cost three times as much as sending one. Batch your questions and instructions into a single prompt to minimize base request charges.