> For the complete documentation index, see [llms.txt](https://docs.cherryai.com.cn/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cherryai.com.cn/docs/en-us/advanced-basic/agent-workspace/models-image.md).

# Model Roles and Agent Drawing

An Agent can use three text-model roles and call the \[Generate Image] tool through the global image-generation model. First, get the task working with a simple setup, then split the model roles as needed.

## Three model roles

| Role        | Main responsibility                                           | Suggested starting point                                                       |
| ----------- | ------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| Main model  | Understand the goal, call tools, and generate the main result | Choose a model that has been verified to use tools reliably                    |
| Plan model  | Break down complex tasks and make plans                       | Start with the same as the main model; split only if it is truly needed        |
| Small model | Simple judgments, classification, and formatting              | Start with the same as the main model to avoid introducing variables too early |

Configuration path: left navigation \[Work] → open the Agent menu → \[Edit] → \[Basics].

{% hint style="warning" %}
When switching models for an in-progress task, the continuity of subsequent replies may change. Replies that have already started generating will not be interrupted by configuration changes; the new model will be used starting from the next message.
{% endhint %}

## Let the Agent generate images directly

Agent drawing uses Cherry Studio's global image-generation model, not the Agent's main model.

{% stepper %}
{% step %}

### 1. Configure the drawing model

Open \[Settings] → \[Default Model] → \[Drawing Model], and select an image-generation model that is already available in the model service.
{% endstep %}

{% step %}

### 2. Check the Agent's image tool

Open the Agent edit window → \[Built-in Tools], and confirm that \[Generate Image] is enabled.
{% endstep %}

{% step %}

### 3. Describe the image in \[Work]

Tell the Agent the image's purpose, subject, style, composition, aspect ratio, and what you do not want to appear. For example:

```
Generate a 16:9 hero image for this product launch copy. The scene is an AI desktop workspace at night, in blue-purple tones, leaving space on the right for the title, with no text or brand logos.
```

{% endstep %}

{% step %}

### 4. Check tool calls and results

Image generation may require tool approval and may also incur provider usage charges. Confirm the model and parameters being called before allowing execution; the result will be shown directly in the Agent task.
{% endstep %}
{% endstepper %}

<figure><img src="/files/b127156db3f9cbfffb53b86f8bf2508d028d0c8a" alt="默认模型设置中的绘画模型选择"><figcaption><p>After selecting the drawing model in the default model, the Agent can use [Generate Image].</p></figcaption></figure>

## Common phrasing

| Goal                              | Prompts should include                         | Example                                                                       |
| --------------------------------- | ---------------------------------------------- | ----------------------------------------------------------------------------- |
| Illustration for an article       | Topic, whitespace, color tone, aspect ratio    | "Landscape 16:9, main subject on the left, title area on the right"           |
| Product sketch                    | Perspective, material, background, constraints | "Isometric view, white background, no text"                                   |
| Social media image                | Platform ratio, focal point, text requirements | "Square, centered subject, no text generated"                                 |
| Modify based on a reference image | Attachment, parts to keep, and parts to change | "Keep the person and composition, only replace the background and color tone" |

<figure><img src="/files/f0cb9ea37113b8f9a19c80cd71df49cbe4423b65" alt="Agent 对话中启用生成图片工具并选择绘画模型"><figcaption><p>After the drawing model is configured, the Agent can generate images directly in tasks; enter [Drawing] only when you need templated batch creation.</p></figcaption></figure>

## Recommended choice

| Need                                            | Recommended approach                                        | Applicable scenario                                | Notes                                                     |
| ----------------------------------------------- | ----------------------------------------------------------- | -------------------------------------------------- | --------------------------------------------------------- |
| Occasionally add an image to a document         | Enable \[Generate Image] in the Agent                       | Illustrations for reports, proposals, and notes    | First confirm the size, style, and purpose                |
| Repeatedly use the same set of image parameters | Use the \[Drawing] template                                 | Brand images, series covers                        | Verify the result once before saving the template         |
| Edit an existing image                          | Provide a reference image and change requests in \[Drawing] | Change colors, expand the image, local adjustments | Keep the original image to avoid irreversible overwriting |

## User case: one release note with three images

The content operations team put the release note into the work directory and had the Agent first extract three visual themes, then generate a hero image, a feature diagram, and a square social media image separately. Only one image was generated each time and the direction was confirmed, avoiding multiple calls in one go and having to redo everything.

<details>

<summary>Why is [Generate Image] unavailable?</summary>

First check whether a drawing model has been selected under \[Settings] → \[Default Model], then confirm that the corresponding provider is enabled and connected properly. Finally, check whether the Agent's \[Generate Image] tool is turned off.

</details>

<details>

<summary>The main model itself can draw images—do I still need a drawing model?</summary>

The Agent's \[Generate Image] tool uses the global drawing model. Even if the main model supports multimodal understanding, it cannot replace the drawing model configuration here.

</details>


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.cherryai.com.cn/docs/en-us/advanced-basic/agent-workspace/models-image.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.