Fireworks AI is an American startup founded in 2022 by Lin Qiao, who previously led PyTorch development at Meta, and others. Headquartered in Redwood City, California, it operates an inference platform that serves open-source large language models and image models through an API, with a proprietary inference engine focused on serving speed and cost efficiency, along with fine-tuning services.
Developers obtain API keys to call models hosted by Fireworks AI or to deploy their own models. It is one of the platforms commonly used to build AI applications that access models from multiple providers through a single interface.