fal MCP Server
Connect to the fal MCP server to search fal.ai models and generate images, video, audio, speech, and 3D from text using AI agents on Gumloop, Claude, or Cursor.
MCP server URL
https://mcp.gumloop.com/fal/mcp
Works with
Tools 4
- 3 credits
Search Models
Search available fal.ai models by category
- 3 credits
Get Model Schema
Get the OpenAPI input/output schema for a fal.ai model
- 110 credits
Run Model
Submit a model inference request to fal.ai. Returns a request_id for polling with get_result.
- 3 credits
Get Result
Poll the status and result of a fal.ai model request
Frequently asked questions
Related MCP servers
Google Drive
Search and access files in Google Drive
- Google Workspace
- OAuth
Google Cloud Storage
Interact with Google Cloud Storage to manage buckets, upload/download files, copy/move objects, and control file access
- Google Workspace
- API key
Google Slides
Create, read, and manage Google Slides presentations
- Google Workspace
- OAuth
Gamma
Interact with Gamma API to create AI-powered presentations, documents, and social media content
- Productivity & Collaboration
- OAuth or API key
Slack
Interact with Slack channels and messages
- Productivity & Collaboration
- OAuth
Notion
Server for interacting with Notion
- Productivity & Collaboration
- OAuth
Further reading
What is fal MCP?
The fal MCP server gives AI agents a way to run generative models on fal.ai. That means agents can search the fal.ai model catalog, inspect a model’s input schema, submit a text prompt for inference, and retrieve the result, across image, video, audio, speech, and 3D generation. It connects to fal.ai’s hosted models, so an agent can generate media on demand instead of you wiring up the fal.ai API yourself.
Generating media through a raw API is fiddly: you have to find the right model, figure out its exact input format, submit the job, and poll for the result. The fal MCP server hands that work to an AI agent that can pick a model, read its schema, run the prompt, and bring back the output. Describe what you want to generate, and your AI agent will run the model for you.
MCP stands for Model Context Protocol, an open standard that lets AI agents call external tools. Normally, using fal.ai means signing up for an API key and writing code against its queue API. Gumloop wraps that behind a hosted server with fal access built in. After you connect, you can search models and generate images, video, audio, speech, and 3D just by chatting with your AI agent.
What you can do with fal MCP on Gumloop
Search the fal.ai model catalog
Find models by generation type, image, video, audio, speech, or 3D, with an optional free-text query, so an agent can pick the right one.
Inspect a model’s input schema
Pull a model’s OpenAPI input and output schema so an agent knows exactly which parameters it accepts before running it.
Generate images from text
Submit a text-to-image prompt to models like Flux or SDXL and get the generated image back.
Generate video, audio, speech, and 3D from text
Submit text-to-video, text-to-audio, text-to-speech, or text-to-3D prompts to the matching models.
Run inference asynchronously
Submit a request to the fal.ai queue and get a request ID, then poll for the result, which keeps long-running video and 3D jobs from timing out.
Let an agent choose the model
Chain search and schema inspection so an agent can browse models, read their inputs, and run the best fit for the task.
How to connect the Gumloop fal MCP Server
Create a Gumloop account
Sign up at gumloop.com. Every new account starts with a 14-day free trial.
Add the fal MCP server
Copy your MCP server URL from Gumloop and add it to your preferred client (Claude, Cursor, or Gumloop workflows). You'll authorize on first use.
Start using fal in your AI workflows
That's it. Your AI agent can now search models and generate images, video, audio, speech, and 3D. Use it inside a Gumloop automation, in Claude Desktop, or in Cursor.
fal MCP use cases
On-demand images for marketing teams
An agent takes a campaign brief, generates hero images with a text-to-image model, and drops the results into Slack or Google Drive for review.
Short video clips for social teams
An agent submits a text-to-video prompt, polls until the clip is ready, and posts the link to the channel handling that campaign.
Voiceovers for content teams
An agent turns a script into speech with a text-to-speech model, then attaches the audio to the draft it is working on.
3D assets for design and product teams
An agent generates a 3D asset from a text prompt so designers have a starting point to iterate on.
Agent-driven model selection
Given a goal, an agent searches the catalog, inspects each candidate model’s schema, and runs the one that best fits the prompt, without a human picking the endpoint by hand.
Why use Gumloop for fal MCP
No API key to manage
Using fal.ai directly means signing up for an API key and wiring it into your code. With Gumloop the fal connection is built in, so there is nothing to generate or store. Just connect and go, or add your own fal API key if you prefer to use your own account and billing.
Works with multiple MCP clients
Use the same hosted fal MCP server in Gumloop, Claude, Cursor, and other MCP-compatible clients. Same server URL, works with any MCP client.
Chain fal with 100+ integrations
An agent can generate an image or clip with fal, then store it in Google Drive, post it to Slack, or drop it into a doc, all in a single run.
Enterprise-grade hosting
Gumloop hosts the server for you, and generated media stays on fal.ai as hosted URLs rather than being kept in Gumloop storage. See trust.gumloop.com for security details.
Start with a free trial
Gumloop offers a 14-day free Pro trial so you can try the fal MCP server before you commit. Paid plans start at $37/month.