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

# AI

> Integrate powerful AI models into your project

The **AI** in Lumi’s Toolbox allows you to seamlessly embed **large language models (LLMs)**  into your own projects. Through natural-language prompts, you can easily activate and configure AI capabilities without complex setup steps — making model integration **faster, smarter, and more intuitive**.

## Supported AI Models

Lumi currently supports multiple models from **Google Gemini** and **OpenAI GPT** families. Each model has different performance and pricing tiers to fit diverse use cases:

| **Model**                               | **Tier**           | **Recommended Use**                                                |
| :-------------------------------------- | :----------------- | :----------------------------------------------------------------- |
| **Gemini 2.5 Flash**                    | Mid-tier (Default) | General-purpose assistant for balanced reasoning and efficiency    |
| **Gemini 2.5 Pro**                      | High-tier          | Advanced reasoning and complex analysis                            |
| **Gemini 2.5 Flash Lite**               | Entry-tier         | Large-scale, low-latency tasks such as text classification         |
| **Gemini 2.5 Flash Image（Nano Banana）** | Mid-tier (Visual)  | Image generation and creative design tasks                         |
| **Gemini 3 Pro Image（Nano Banana Pro）** | High-tier (Visual) | Create and edit images with high level of precision and control    |
| **GPT-5**                               | High-tier          | Complex reasoning and decision-making applications                 |
| **GPT-5 Mini**                          | Mid-tier           | Moderate-complexity workflows and automation tasks                 |
| **GPT-5 Nano**                          | Entry-tier         | High-volume, lightweight tasks                                     |
| **GPT-4o**                              | High-tier          | Advanced multimodal understanding and high-quality text generation |

<Tip>
  You can mix and match models in one project — for example, using Gemini for fast responses and GPT for creative generation.
</Tip>

## Enable AI Models

By default, AI capabilities are **disabled**. You can activate them in one of three ways:

1. Via **conversation** — when you ask to integrate or use AI features, Lumi will prompt you to enable AI.

   <img src="https://mintcdn.com/blockdance/shlU-5w8u3irmHaO/images/ai1.png?fit=max&auto=format&n=shlU-5w8u3irmHaO&q=85&s=370df09e4b9a92c8af93be05ca6935a5" alt="Ai1 Pn" width="1280" height="987" data-path="images/ai1.png" />
2. From the **Toolbox** on the homepage

   <img src="https://mintcdn.com/blockdance/shlU-5w8u3irmHaO/images/ai2.png?fit=max&auto=format&n=shlU-5w8u3irmHaO&q=85&s=7cb208cafb908943f68d2c0acc2b938c" alt="Ai2 Pn" width="1280" height="944" data-path="images/ai2.png" />
3. From your **workspace**

   <img src="https://mintcdn.com/blockdance/xssS1okt2OxC-BM9/images/3.png?fit=max&auto=format&n=xssS1okt2OxC-BM9&q=85&s=5134df4dcfcce1e7ae28c90c8b16df66" alt="3 Pn" width="1080" height="606" data-path="images/3.png" />

<Tip>
  When you enable AI without specific requirements, Lumi automatically integrates the **default model (Gemini 2.5 Flash)**. You only need to request it in the conversation to switch models.
</Tip>

## Manage AI Models

A dedicated **AI** panel is now available in your workspace, providing full visibility of your AI usage.

<img src="https://mintcdn.com/blockdance/xssS1okt2OxC-BM9/images/4.png?fit=max&auto=format&n=xssS1okt2OxC-BM9&q=85&s=9e01c6a7b204c73c1c32a9141039b96f" alt="4 Pn" width="1090" height="1124" data-path="images/4.png" />

Here, you can:

* Monitor real-time model calls and usage, with filtering options by time range.
* View detailed Tool Credit spending for each integrated model.

## Subscription & Billing

* No subscription requirement — all Lumi users can access AI integration.
* AI usage is billed together with **Tool Credit**.
* All AI usage within the project is charged to the creator's Tool Credit balance, and project users bear no costs.