fal Launches MCP Server Connecting AI Agents to 1,000+ Generative Models

Mar 20, 2026 · Updated Apr 25, 2026

fal, a generative AI model hosting platform, launched the fal MCP Server — a hosted endpoint connecting MCP-compatible AI assistants to 1,000+ generative AI models across image, video, audio, and 3D. The server exposes 9 tools — search_models, run_model, submit_job, and recommend_model among them — that the assistant selects and calls automatically. It's stateless, hosted on Vercel, and API keys are never stored server-side.

This removes the manual step of writing API calls against generative AI services. The assistant discovers models, runs them, chains them across modalities, and manages job queues — all in one conversation turn. Setup works with Claude Code, Claude Desktop, Cursor, and Windsurf.

Connect your AI assistant to fal MCP and run a multi-model pipeline — have it write a prompt, generate an image, then convert to video — in one natural-language request. The server is free; you pay for model runs at standard fal pricing.

View the full update on blog.fal.ai

fal

@falMar 19

🔥 The fal MCP Server is live ! Connect Claude, Cursor, or any AI assistant to 1,000+ generative AI models. Search models, generate images, create videos, check doc, create app : from a conversation. All the link to access it here 👇 https://t.co/z1iGv7W6bf

View on X