Sign in
Ideogram MCP

Add Ideogram to any AI agent

Ideogram MCP is Ideogram's official Model Context Protocol server. It lets Claude, ChatGPT, Cursor or any assistant that supports MCP work in your Ideogram account: it can generate images, SVG artwork and video, edit and reframe images, remove backgrounds, upscale results, save them to collections, and start custom model training. You connect once with the server URL and sign in with your Ideogram account through OAuth, so there is no API key to manage.

You can sign in with any Ideogram account. Generation spends your workspace's credits.
Claude Connected to Ideogram MCP

Install

Connect Claude, ChatGPT, Cursor or any MCP client.

The server speaks Streamable HTTP at https://mcp.ideogram.ai/mcp. Authentication is OAuth through your Ideogram account. The first time a client connects, a browser window opens for sign-in. In Claude, you add Ideogram as a custom connector with that URL.

1. Open Claude and go to Customize → Connectors.
2. Select + Add, then Add custom connector.
3. Name it Ideogram and paste the server URL:

   https://mcp.ideogram.ai/mcp

4. Finish the setup, then sign in to Ideogram
   when the browser opens.

Custom connectors work in Claude on the web, Claude Desktop and Cowork. On Team and Enterprise plans, an Owner adds the connector once under Organization settings → Connectors, then each member connects.

claude mcp add --transport http ideogram https://mcp.ideogram.ai/mcp

Run it in your terminal. Then type /mcp in Claude Code, pick ideogram and sign in to Ideogram in the browser.

1. In ChatGPT, open Plugins.
2. Select the plus button, then Add custom MCP server.
3. Name it Ideogram and enter the server URL:

   https://mcp.ideogram.ai/mcp

4. Set authentication to OAuth, accept the notice
   and select Create as a plugin.
5. Sign in to Ideogram when ChatGPT asks.

OpenAI documents custom MCP servers for ChatGPT Business, Enterprise and Edu workspaces on the web, where an admin or owner adds the server.

{
  "mcpServers": {
    "ideogram": {
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Open Cursor Settings → MCP → Add new server and paste this into mcp.json. Cursor asks you to sign in to Ideogram on first use.

{
  "mcpServers": {
    "ideogram": {
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Cline reads the same JSON shape from cline_mcp_settings.json.

{
  "mcp": {
    "ideogram": {
      "type": "remote",
      "url": "https://mcp.ideogram.ai/mcp",
      "enabled": true
    }
  }
}

Add to ~/.config/opencode/config.json or a project-local opencode.json.

{
  "mcpServers": {
    "ideogram": {
      "transport": "streamable-http",
      "url": "https://mcp.ideogram.ai/mcp"
    }
  }
}

Streamable-HTTP transport. Refer to your Hermes build's config path; the JSON shape follows the standard mcpServers convention.

POST https://mcp.ideogram.ai/mcp
Authorization: Bearer <oauth-token>
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "method": "tools/list",
  "id": 1
}

Any client that speaks MCP streamable-HTTP works. Discover OAuth via /.well-known/oauth-authorization-server.

Capabilities

Ask in your own words, and the assistant picks the tool.

Once connected, the assistant sees Ideogram's tools and chooses among them from your request. Generation and training run as background jobs, so a long batch does not block the conversation.

Area What the assistant can do Example request
Generate Make images with Ideogram 4.5, 4.0 and 3.0, including transparent backgrounds, your custom models, and batches of up to 500 prompts in one request. Make SVG artwork with Ideogram Vector, and generate video. "Make 12 Instagram posts for our fall sale, same font and colors."
Edit Edit, remix, upscale and reframe images. Remove or replace a background, remove objects, resize an ad for a new placement, vectorize an image, and change colorways or materials on a product photo. "Turn this banner into a 9:16 story and upscale it."
Understand Describe an existing image as a prompt, then change part of the description and generate from it. "Describe this poster, then make one like it for spring."
Organize List your recent images, create and rename collections, and add or move images between them. "Put these in a collection called Fall Campaign."
Custom models Create a training dataset, upload reference images, start a training run, and list the models you can generate with. "Train a model on these 20 product photos."

In practice

Try one of these to start.

Drop these into your assistant after connecting. The agent picks the right tool from the catalog.

Workflows

Generate, edit, and organize assets in one agent workflow.

Each workflow runs in a single conversation. The agent chains the tools in order, and you review the results before anything ships.

Asset packs from a brief

Hand the agent a one-paragraph campaign brief. It returns a hero, four social cards, and two print posters in a single conversation, saved into a named collection. You review the set in the collection and choose which images to upscale.

  • generate_images
  • manage_collections
  • edit_images

Catalog cleanup

Feed product photos into the chat. The agent removes backgrounds and reframes them for marketplace ratios. You check each cutout before the agent saves the cleaned assets to a shared collection.

  • upload_image
  • edit_images
  • manage_collections

Train and iterate

Drop a folder of brand references. The agent creates a dataset, uploads the assets, and starts a training run. The run finishes later, and you generate with the model once training completes.

  • manage_dataset
  • train_model
  • get_models
Build the same workflow through the API

Sign-in and credits

The assistant works in your account and spends your credits.

Sign-in

You sign in with your Ideogram account in the browser, using any sign-in method Ideogram offers. No API key is created or stored in the client.

Workspace

If you belong to more than one workspace, you choose one when you first connect. Every client you connect uses that workspace, and you can ask the assistant to switch it later.

Credits

Requests spend the chosen workspace's credits: your plan's monthly credits first, then credits you bought. When they run out, the assistant returns a link to add credits or upgrade, and requests do not fall back to the slower queue the app offers.

Limits

Files sent through the assistant can be up to 50 MB, and one request can submit up to 500 prompts. The number of jobs that run at once follows your plan.

See plans and credits

MCP or API

Use MCP in a conversation and the API in your own software.

Ideogram MCP server Ideogram API
Use it when A person works with an assistant in a chat. Your own software or pipeline makes the requests.
Sign-in OAuth with your Ideogram account. An API key in the Api-Key header.
Billing Your workspace's credits. API usage, billed to the key's workspace.
Results They appear in the chat and in your Ideogram account. Your code downloads them from the response.

Questions

Frequently asked

What is the Ideogram MCP server?
The Ideogram MCP server is Ideogram's official Model Context Protocol server, hosted at https://mcp.ideogram.ai/mcp. It lets an AI assistant such as Claude, ChatGPT or Cursor generate, edit and organize images, SVG artwork and video in your Ideogram account from a conversation.
Can Claude generate images with Ideogram?
Yes. Add Ideogram to Claude as a custom connector with the server URL https://mcp.ideogram.ai/mcp and sign in to your Ideogram account. Claude can then generate, edit, upscale and reframe images with Ideogram's models inside the conversation, and the results appear in your Ideogram account.
How do I connect Ideogram's MCP server to Claude or Cursor?
In Claude, open Customize, then Connectors, select Add custom connector, paste https://mcp.ideogram.ai/mcp and sign in to Ideogram. In Cursor, open Cursor Settings, then MCP, add a new server and paste the Ideogram entry into mcp.json. In Claude Code, run claude mcp add --transport http ideogram https://mcp.ideogram.ai/mcp.
Which clients are supported?
Any client that speaks the MCP Streamable HTTP transport works. That includes Claude, Claude Code, ChatGPT, Cursor, Cline and OpenCode, and any custom client that implements the protocol.
How does authentication work?
Your client opens a browser window and you sign in to your Ideogram account through OAuth, with any sign-in method Ideogram offers. The server then acts as you, so no API key is created or stored in the client.
Does the Ideogram MCP server use my plan's credits?
Yes. Requests spend the credits of the workspace you chose when you connected: your plan's monthly credits first, then credits you bought. When the credits run out, requests do not fall back to the slower queue that the Ideogram app offers. The assistant returns a link to add credits or upgrade, and you can retry after that. Ideogram pricing lists what each plan includes.
How does a Claude Team or Enterprise organization add Ideogram?
An Owner adds the Ideogram server URL once as a custom connector under Organization settings, then Connectors. Each member then opens Customize, then Connectors, selects Connect and signs in with their own Ideogram account, so every person works with their own access and workspace.
What can an assistant do through the Ideogram MCP server?
It can generate images, SVG artwork and video. It can edit, remix, upscale and reframe images, remove or replace backgrounds, remove objects, resize ads and vectorize images. It can also describe an image as a prompt, manage collections, and train and use custom models.
What are the limits of the Ideogram MCP server?
Files sent through the assistant can be up to 50 MB, and one request can submit a batch of up to 500 prompts. The number of jobs that run at once follows your plan, as it does in the Ideogram app.
When should I use the Ideogram API instead of the MCP server?
The MCP server is for a person working with an assistant, signed in with their own Ideogram account. To build Ideogram into your own software or run server-to-server pipelines, call the Ideogram API with an API key.
Is the docs MCP server the same thing?
No. The server at https://developer.ideogram.ai/_mcp/server only searches the API documentation and cannot generate images. The server on this page, https://mcp.ideogram.ai/mcp, is the one that generates and edits.