Skip to Content
Setup GuidesChatGPT

How to Use Blender MCP with ChatGPT

Blender MCP can connect to ChatGPT, but support is still evolving. ChatGPT Desktop has added MCP server support, and the setup is similar to Claude and Cursor. If you hit limits with ChatGPT’s MCP integration, we recommend using Claude or Cursor, which have mature MCP support and are well-tested with Blender MCP.

Current Status: ChatGPT MCP Support

ChatGPT Desktop now supports the Model Context Protocol (MCP), but the experience is not yet as stable as with Claude or Cursor. Typical options are:

  1. ChatGPT Desktop + mcp.run — Some users connect MCP tools via mcp.run , which acts as a gateway for ChatGPT Desktop.
  2. ChatGPT Apps (workspace) — Organization admins can add MCP servers through workspace settings in the ChatGPT Apps (beta) program.
  3. OpenAI Agents SDK — Developers can build custom agents with MCP support using the OpenAI Agents SDK (Python/JavaScript), with transport options including stdio and SSE.

If direct MCP setup in ChatGPT Desktop is limited in your region or plan, Claude Desktop or Cursor with Blender MCP remains the most reliable option. See our Claude setup guide for a step-by-step walkthrough.

Prerequisites

Step 1: Install uv Package Manager

The Blender MCP server runs via uvx. Install uv first:

Do not skip this step. The MCP server will not work without uv.

macOS:

brew install uv

Windows (PowerShell):

powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

Add uv to PATH: $env:USERPROFILE\.local\bin

Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Verify: uv --version

Step 2: Install the Blender Addon

Download addon.py

Get the latest addon.py from the Blender MCP GitHub repo .

Install in Blender

Open Blender → Edit → Preferences → Add-ons → “Install…” → Select addon.py.

Enable it

Check the box next to “Interface: Blender MCP”.

Step 3: Configure MCP in ChatGPT Desktop

ChatGPT Desktop’s MCP configuration varies by version and region. In general:

If ChatGPT Desktop has built-in MCP settings:

  1. Open ChatGPT Desktop → Settings (or Preferences)
  2. Look for Developer, Integrations, or MCP options
  3. Add a new MCP server with:
    • Command: uvx
    • Args: blender-mcp

The config format (when supported) is:

{
  "mcpServers": {
    "blender": {
      "command": "uvx",
      "args": ["blender-mcp"]
    }
  }
}

If using mcp.run:

  1. Sign into mcp.run  from ChatGPT Desktop (if integration is available)
  2. Create or select a tool profile
  3. Add the Blender MCP server if it’s offered in the catalog

ChatGPT MCP support is still evolving. If you don’t see MCP options in your ChatGPT Desktop, OpenAI may not yet have rolled them out for your plan or region. In that case, use Claude or Cursor instead — both have stable, well-documented Blender MCP support.

Step 4: Start the Connection

Start the Blender server

In Blender’s 3D Viewport, press N → find “BlenderMCP” → click “Connect to Claude” (works for all MCP clients).

Verify MCP tools in ChatGPT

If MCP is connected, you should see Blender-related tools available in the chat. Try: “Create a red cube in Blender”.

Workarounds If ChatGPT MCP Is Limited

If ChatGPT Desktop doesn’t yet support MCP in your setup:

  1. Use Claude — Claude setup guide. Native MCP support and widely used with Blender MCP.
  2. Use Cursor — Cursor setup guide. IDE with built-in MCP support.
  3. Use VSCode + Cline — VSCode setup guide. Cline adds MCP support to VSCode.
  4. OpenAI API + custom agent — The OpenAI Agents SDK supports MCP. You can build a custom agent that connects to Blender MCP, but this requires development work.

Troubleshooting

MCP options not visible in ChatGPT

ChatGPT Desktop MCP may not be available for your account or region. Use Claude or Cursor instead.

”Connection refused” or server not found

  • Ensure the Blender addon is running (BlenderMCP panel, Connect clicked)
  • Restart ChatGPT Desktop after config changes
  • Verify uv: uv --version
  • Don’t run uvx blender-mcp manually — the client should start it

First command fails

Known quirk: the first command sometimes fails. Send a second message and retry.

Is Blender MCP Free with ChatGPT?

Yes. Blender MCP is free and open source (MIT License ). Blender is free. Costs come from your ChatGPT subscription; Blender MCP itself has no additional fee.

For the best Blender MCP experience today, we recommend Claude or Cursor. As ChatGPT’s MCP support matures, setup will likely become simpler.

For a technical deep dive into how the server works, see our MCP Server architecture guide.

Using a different AI client?

Claude Setup →Cursor Setup →VSCode Setup →Gemini Setup →Ollama Setup →

FAQ

Does ChatGPT support Blender MCP?

ChatGPT Desktop has added MCP server support, but it is not yet as mature as Claude or Cursor. Configuration options vary by version and region. For the most reliable Blender MCP experience, we recommend Claude or Cursor.

How do I add MCP servers to ChatGPT?

ChatGPT Desktop may offer MCP settings under Developer, Integrations, or similar. Some users connect via mcp.run. If MCP options are not visible, OpenAI may not have enabled them for your account yet. Use Claude or Cursor as alternatives.

Can I use Blender MCP with the OpenAI API?

Yes. The OpenAI Agents SDK supports MCP and multiple transports (stdio, SSE, etc.). You can build a custom agent that uses Blender MCP, but this requires development. For a ready-to-use setup, Claude or Cursor are easier.

What's the best AI for Blender MCP?

Claude and Cursor currently offer the most stable and documented Blender MCP integration. ChatGPT support is emerging. See our Claude and Cursor setup guides for step-by-step instructions.

Get Started with Blender MCP

Blender MCP is free, open-source, and community-driven. Star the repo, report issues, or contribute — all on GitHub.

View on GitHub →
Last updated on
VSCodeGemini