- MODEL SEARCH
- JOB STATUS
- OUTPUT DATA
- HUMAN INPUT
Short answer
How can I run machine learning models through my AI?
Your AI searches the Replicate library for the specific model you need, selects a version, and starts the prediction job. You get the output directly in your conversation. This breakdown shows how the AI handles the model selection and execution steps.
The outcomes
What you get from your request
Depending on what you ask, your AI will produce one of these outcomes.
MODEL SEARCH
Found specific models
The AI identifies relevant models in the Replicate library based on your description. You get a list of options to choose from.
JOB STATUS
Live prediction tracking
The AI monitors the progress of your inference job. You'll know exactly when the model finishes processing your input.
OUTPUT DATA
Direct model results
Once the job finishes, the AI pulls the prediction data. You receive the final image, text, or file directly in your chat.
HUMAN INPUT
Refining the model
If the model isn't what you expected, you decide which version to use or how to adjust your prompt. The AI waits for your direction.
The workflow
What your AI does when <the thing arrives>.
The AI handles the technical handshakes with Replicate so you can focus on the prompt.
Locate the model
The AI scans the Replicate library to find the exact model or type of model you described.
search_modelsSelect a version
After finding a model, the AI identifies the specific version or configuration required for your task.
get_model_versionStart the job
The AI sends your input data to Replicate to begin the inference process.
create_predictionFetch the result
The AI checks the status of the running job and retrieves the final output once it's ready.
get_prediction
Try it
Copy these to start.
Use these prompts to trigger your first Replicate jobs.
Starting points
Swap out the model names and input descriptions to match what you actually want to create.
Replicate Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_ULeVbcRjoLCkIr9x93q2HML8VJd4AXmk4CWHHPso/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Replicate capabilities are ready to use.
{
"mcpServers": {
"replicate-extended-mcp": {
"url": "https://edge.vinkius.com/vk_preview_ULeVbcRjoLCkIr9x93q2HML8VJd4AXmk4CWHHPso/mcp"
}
}
}Find a model in Replicate that can upscale low-resolution images.
Search for a text-to-image model and run a prediction with the prompt 'a neon cyberpunk city'.
Use the Stable Diffusion model to generate an image of a cat wearing a space helmet.
Look for a model that can remove backgrounds from photos and run it on my uploaded file.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Replicate once, then ask.
One click connects this MCP to your client. Your credentials stay encrypted and secure on our side.
Connect Replicate to your AIFAQ
How this task behaves
- 01
Can the AI create new models in my Replicate account?
No. The AI can only search for existing models and run predictions on them.
- 02
Does the AI handle the billing for Replicate?
No. You manage your own Replicate account and billing; the AI simply triggers the jobs through your connection.
- 03
Can I use this to run any model on Replicate?
Yes, as long as the model is available in the Replicate library, the AI can find and use it.
- 04
Will the AI automatically retry if a prediction fails?
No. If a prediction fails, the AI will report the error to you, and you must decide whether to try again.
- 05
Can the AI delete models I have uploaded?
No. This MCP does not have the capability to delete or modify your existing models or account settings.
More questions about Replicate? See everything the Replicate Connector can do
Connect Replicate to Claude, Cursor, ChatGPT & more
