fal

fal builds a generative media cloud platform that gives developers access to more than 600 production-ready image, video, audio and 3D models through a single API. The company was founded in 2021 and operates worldwide, serving over 1,000,000 developers and, through enterprise customers, hundreds of millions of end users. Its inference engine is billed as up to 10x faster than alternatives, with particularly fast paths for models such as SDXL and Whisper.

Underneath the model catalogue sits a serverless GPU infrastructure running on recent NVIDIA hardware, including H100, H200 and B200 chips, alongside dedicated compute clusters. The platform supports on-demand and reserved capacity priced from $1.2 per hour for premium GPUs, and is engineered to scale from early prototypes to more than 100m daily inference calls at 99.99% uptime. Alongside inference, fal provides tooling for training and deploying custom AI models at scale.

The engineering profile is oriented around speed and infrastructure: machine learning inference, serverless GPU computing, cloud infrastructure and model deployment are the core technical domains. The company reports SOC 2 compliance and an emphasis on enterprise-grade security, alongside a developer-first approach to its products.