Featherless AI

Featherless provides serverless inference for more than 40,000 open-weight AI models through a single OpenAI-compatible API, with dedicated GPU capacity also offered. The website lists chat, agent and dedicated-GPU plans, including flat monthly chat subscriptions, per-token billing for developer and production API use, and annual contracts for dedicated H100, MI325X, B200 and B300 hardware.

Customers named on the site include chat and roleplay users, developers of coding agents and automation tools (such as Cline, OpenHands and n8n), and enterprises that need dedicated GPUs and private fine-tunes. Revenue comes from monthly subscriptions, metered token usage, credit top-ups and custom contracts.

The company was founded in 2023 by Eugene Cheah, Harrison Vanderbyl and Wesley George, and was previously known as Recursal.AI. Its headquarters is at 717 Market Street, San Francisco, with satellite offices in Toronto and Singapore and a presence in Europe. The website reports a $20M Series A in 2026, co-led by AMD Ventures and Airbus Ventures. Open roles cover go-to-market in the US, Canada and Europe, AI research and machine learning engineering, and a Chief of Staff position.

Product: Serverless LLM inference API for 40,000+ open-weight models, plus chat subscriptions and dedicated GPU capacity