• INFRASTRUCTURE
  • HARDWARE
  • PREDICTION
  • HUMAN

How to manage model deployments with your AI

Connect Replicate to your AI client to manage production infrastructure. You'll move from testing models to running them on dedicated hardware through simple chat commands.

Ask AI about this page

Short answer

How can I manage model deployments with my AI?

You can spin up private deployments and select specific hardware configurations directly through your AI client. This gives you control over the compute resources your models use. You'll see how the AI handles the setup in the breakdown below.

Deployment outcomes

What you get from your AI

These are the results of your deployment commands.

INFRASTRUCTURE

Private deployments

Your AI sets up dedicated instances for your models. You get isolated environments that don't share resources with public endpoints.

HARDWARE

Custom compute settings

The AI checks available hardware and configures your deployment to match your needs. You get the exact GPU or CPU setup required for your workload.

PREDICTION

Live model testing

Once a deployment is live, your AI can run predictions immediately. This lets you verify that the hardware handles your model's specific requirements.

HUMAN

Manual cost review

The AI executes the technical setup, but you must still monitor your usage and costs. You remain in control of the budget and final resource decisions.

The workflow

What your AI does when a deployment request arrives.

The AI acts as the bridge between your chat and the Replicate infrastructure.

  1. Check hardware

    The AI looks at what hardware is currently available to ensure your requested configuration can actually run.

    list_hardware

  2. Provision deployment

    Your AI creates a private, dedicated instance of your model with the specific settings you requested.

    create_deployment

  3. Run test prediction

    The AI sends data to your new deployment to confirm everything is working as expected.

    create_deployment_prediction

  4. Verify setup

    The AI uses the deployment tool to confirm the instance is active and ready for use.

    create_deployment

Try it

Copy these to start.

Use these prompts to begin managing your infrastructure.

Starting points

These are opening lines. Swap out the model names and hardware types for your specific needs.

Replicate Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_ULeVbcRjoLCkIr9x93q2HML8VJd4AXmk4CWHHPso/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Replicate capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "replicate-extended-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_ULeVbcRjoLCkIr9x93q2HML8VJd4AXmk4CWHHPso/mcp"
    }
  }
}
Copy into chat04

  • List the available hardware options so I can choose a configuration.

  • Create a private deployment for the stability-ai/sdxl model using an A100 GPU.

  • Run a prediction on my latest deployment to make sure the hardware is working.

  • Check what hardware is available for a high-memory deployment.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Replicate once, then ask.

Link your account once and your credentials stay encrypted. You can then manage your entire deployment workflow from your preferred AI client.

Connect Replicate to your AI

FAQ

How this task behaves

  • 01

    Can the AI delete my existing deployments?

    No. This MCP only has tools to create deployments, list hardware, and run predictions. You must manage deletions manually.

  • 02

    Can the AI change my billing settings?

    No. The AI cannot access or modify your payment methods or billing information.

  • 03

    How does the AI know which GPU to use?

    The AI uses the list_hardware tool to see what is available, then applies your specific request to the create_deployment tool.

  • 04

    Can I use this to run public models?

    Yes, you can create private deployments for models available on the platform.

  • 05

    Does the AI handle the actual model training?

    No. This connector is for managing deployments and running predictions, not for training new models.

  • More questions about Replicate? See everything the Replicate Connector can do

Connect Replicate to Claude, Cursor, ChatGPT & more