How to Use the NVIDIA Vision MCP in Claude

Q: How does this server protect my scanned documents and images?

Every image or scanned document processed by documentqa runs inside an ephemeral, zero-trust V8 Isolate sandbox. Vinkius destroys the sandbox immediately after the API call completes, ensuring your raw visual data is never stored or used for training.

Run local visual analytics and generate assets in Claude Desktop with NVIDIA Vision models.

See Vinkius in Action

Works with every AI agent you already use

…and any MCP-compatible client

MCP Servers - Free for Subscribers

Connect NVIDIA Vision MCP to Claude Desktop

Create your Vinkius account to connect NVIDIA Vision to Claude Desktop and route execution through our secure gateway. The platform manages server hosting, runtime updates, and security layers. Configuration requires no manual server provisioning.

GDPR Free for Subscribers

Setup NVIDIA Vision with Claude Desktop

Ask AI about this MCP

ChatGPT

Claude

Perplexity

Analyze local files in Claude Desktop

The `document_qa` tool extracts text and answers questions about scanned receipts, forms, and document images directly on your machine. This runs locally via the Claude Desktop app using stdio transport, meaning your files don't leave your local system before the MCP Server processes them. Your agent reads the raw visual data to parse tables and key-value pairs without manual data entry. You get direct answers about your invoices right in the chat interface.

Generate visual assets using SDXL

The `generate_image` tool outputs 1024x1024 images using Stable Diffusion XL or Stable Diffusion 3 Medium based on your text prompts. Because Vinkius manages the connection, Claude Desktop handles the API keys while you focus on writing prompts. You specify the model and size, and the server returns the generated asset. This setup bypasses complex local GPU configurations entirely.

Locate objects with NVIDIA Vision MCP Server

The `visual_grounding` tool finds the exact coordinates of specific items or phrases within your images. Claude Desktop uses this MCP tool to pinpoint elements on a webpage screenshot or identify UI components during design reviews. Combine this with `detect_objects` and `image_segmentation` to get a complete breakdown of any visual scene. Your agent handles the multi-step analysis automatically, mapping out every boundary box in your chat window.

Setup guide

Set up NVIDIA Vision MCP in Claude Web or Desktop

1

Open Claude Settings

Go to claude.ai, click your profile icon, then navigate to Customize → Connectors.
2

Add Custom Connector

Click the "+" button and select Add custom connector. Paste your Vinkius endpoint URL: https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp Replace [YOUR_TOKEN_HERE] with your token from cloud.vinkius.com. For OAuth-protected servers, expand Advanced settings to add credentials.
3

Start a conversation

Open a new chat. The NVIDIA Vision MCP tools are available immediately — no restart needed.

Endpoint URL

https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp

No configuration file needed — paste the URL directly in the Claude web interface.

Available on Free (1 connector), Pro, Max, Team, and Enterprise plans.

Prerequisites

Claude Desktop installed (macOS or Windows)
Active Vinkius subscription with a valid endpoint token

1

Open Claude Desktop Settings

Click the menu icon at the top-left corner, go to Settings → Developer → Edit Config. This opens claude_desktop_config.json in your default text editor.
2

Paste the NVIDIA Vision MCP configuration

Copy the JSON snippet on the right into the mcpServers object. Replace [YOUR_TOKEN_HERE] with your endpoint token from cloud.vinkius.com.
3

Restart Claude Desktop

Close and reopen the application. Claude needs a full restart to load new MCPs — refreshing a conversation is not enough.
4

Verify the connection

Open a new conversation. Click the 🔌 icon at the bottom of the message input. You should see tools listed under nvidia-vision-mcp.

json

{
  "mcpServers": {
    "nvidia-vision-mcp": {
      "url": "https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp"
    }
  }
}

Get your connection token →

Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by NVIDIA. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.

Why Choose Vinkius

Vinkius connects your tools to AI with real-time monitoring and automatic cost savings — all from one dashboard.

Connect NVIDIA Vision now

Real-time monitoring

Live

visibility into every interaction

Connect your favorite tools to your AI and see exactly what's happening — every request, every response, in real time.

Built-in savings

60%

lower AI costs

Vinkius compresses data between your apps and your AI automatically. Lower bills every month — no configuration required.

Single dashboard

One

place for every integration

Every tool your AI connects to, managed from a single screen. One account, complete control.

Common questions about NVIDIA Vision MCP in Claude Desktop

Open your configuration file and add the server settings under the `mcpServers` key. Use the command and arguments provided by Vinkius to establish the stdio connection. Restart Claude Desktop, and you will see the new tools available via the hammer icon.

Yes, if you use the Claude Web interface in your browser. You add the remote MCP Server URL in the Integrations settings, and the tools become available across all your browser-connected devices.

You can use the generation tool to call Stable Diffusion 3 Medium or Stable Diffusion XL. Check the current list of available NIM models by running `list_vision_models` directly from your chat.

No. Vinkius runs the NVIDIA Vision MCP Server in a secure, remote V8 Isolate sandbox. Your local machine only handles the Claude Desktop interface, while the heavy image processing occurs on high-performance remote infrastructure.

Every image or scanned document processed by `document_qa` runs inside an ephemeral, zero-trust V8 Isolate sandbox. Vinkius destroys the sandbox immediately after the API call completes, ensuring your raw visual data is never stored or used for training.

Use it with your favorite AI tools

Connect this server to Cursor, Claude, VS Code, and more.

OpenAI Agents SDK sdk-python

Google ADK sdk-python

Pydantic AI sdk-python

Vercel AI SDK sdk-typescript