Fal.ai, founded in 2021 by former Amazon and Coinbase engineers, runs a serverless inference platform for generative media models, letting developers generate images, video, and audio through a single API call without provisioning or managing GPU infrastructure themselves.
That infrastructure focus, rather than building its own models, let Fal.ai ride the growth of the image and video generation models it serves: inference revenue doubled to roughly $400 million, and the company raised $140 million led by Sequoia at a $4.5 billion valuation in December 2025, with a further round targeting an $8 billion valuation reported in March 2026.
Pricing runs usage-based, billed per GPU-second or per model output, with image generation commonly landing around $0.01 to $0.08 per image and video models priced per second of output. For a developer who wants to call state-of-the-art image and video generation models through one API without managing GPU capacity, Fal.ai's serverless inference layer addresses that directly.
