What is Featherless
Featherless provides serverless inference for more than 40,000 open-weight models through one OpenAI-compatible API, with separate plans for human chat, developer traffic and dedicated business capacity.
Chat and Developer
As of September 19, 2026, Chat costs US$25 per month. It includes the full model catalogue, unlimited tokens for person-typed chat, a 32,000-token context allowance and four concurrent units. Featherless says prompts and completions are not logged.
Developer costs US$50 per month in usage credits. It is designed for coding agents and production API traffic, is billed per token, supports context windows up to 256,000 tokens and up to 100 concurrent units, and includes one agent environment. The live catalogue publishes the current input and output rate for each model.
Business
Business uses custom pricing for dedicated H100, MI325X, B200 or B300 capacity. It adds private fine-tunes, no shared rate limits, named engineering support and a service-level agreement. Featherless states that all plans can access its full catalogue, but model compatibility, context and per-token rates vary by model.
Official sources: Featherless current plans, Featherless plan guide, Featherless live model catalogue.