Fal.ai MCP Server is a Model Context Protocol (MCP) server that enables Claude Desktop (and other MCP clients) to generate media using Fal.ai models. It exposes a high-performance, asynchronous API built on fal_client.run_async(), with support for long-running tasks via a queue API that provides progress updates. The server is non-blocking, ensuring responsive behavior as tasks execute in the background. It supports multiple transport modes: STDIO for traditional MCP communication, HTTP/SSE for web-based access via Server-Sent Events, and a dual mode that runs both transports concurrently. This makes it suitable for both CLI workflows and web clients. Supported media capabilities include image generation (flux_schnell, flux_dev, sdxl), video generation (svd, animatediff), and audio generation (musicgen), as well as text-to-speech (bark) and transcription (whisper), image upscaling, and image-to-image transformations. Quick Start covers Docker, PyPI, and source installations, along with Claude Desktop configuration. Documentation references Docker and HTTP transport usage, and the MCP 1.0 specification is acknowledged. The server is designed to simplify the integration of Fal.ai models into Claude Desktop and other MCP clients.