Featherless

Featherless is a serverless AI inference platform for open-weight models via API. Browse models, make chat and completion calls, and build without servers.

AI Developer Tools AI API Design LLM

Visit Website

Serverless LLM hosting for open-weight models

Featherless is a serverless AI inference platform focused on hosting open-weight language models through an API. Its homepage and docs describe a single API key that gives access to a large and continually expanding model library, including popular families such as Qwen, Llama, Mistral, DeepSeek, and RWKV.

The product is aimed at developers who want to call open models without managing their own infrastructure. The documentation covers getting started, API reference material, models and compatibility, concurrency limits, tool calling, vision, embeddings, and application guides, which suggests a platform built for both experimentation and production use.

Features

Serverless model access

Featherless exposes open-weight models through serverless APIs, so users can call models without provisioning or managing their own inference servers.

Large open-model library

The platform presents a large, searchable model library with more than 30,000 models on the homepage and a broader catalog view on the models page.

API endpoints for common LLM tasks

The docs point to common endpoints for listing models, chat, text completions, tokenization, and streaming, which supports a standard API workflow.

Developer documentation

Documentation sections cover quickstart, models and compatibility, plans, concurrency limits, tool calling, vision, embeddings, and application guides.

Catalog filtering and discovery

The models page includes filters for architecture, language, domain, training, license, modality, capability, release recency, and popularity, which helps narrow the catalog.

Tiered plan constraints

The pricing page shows plan-based limits such as model-size access, concurrent connections, and context window caps.

Use cases

Ship an app with hosted model inference
Use Featherless when you want to call open models from an application without running your own inference stack. The platform is positioned around serverless API access and a single API key.
Evaluate and select a model from the catalog
Use the model browser to compare families, capabilities, and release recency before choosing a model for a task such as coding, reasoning, chat, or roleplay.
Prototype against common LLM endpoints
Use the documentation as a starting point for implementing chat, completions, streaming, tokenization, tool calling, vision, or embeddings in a product workflow.
Match usage to an available plan
Use the plan pages to understand whether your workload fits the visible concurrency and context limits before you adopt the service more broadly.
Work with specific open-model families
Use the site’s model pages and docs to explore open-weight model compatibility if you need to support a specific family such as Qwen, Llama, Mistral, DeepSeek, or RWKV.

Pros and Cons

Pros

Provides API-based access to a large catalog of open-weight models.
Supports common LLM workflows such as chat, completions, tokenization, and streaming.
Documents a range of application areas, including tool calling, vision, embeddings, and compatibility guidance.
Uses a model library with filters that make discovery easier across architectures, languages, licenses, and capabilities.
Offers plan-based access with visible concurrency and context-window limits, which gives at least some operational framing before signup.

Cons

The public pricing information is partial, so readers cannot see the full plan structure from the collected source text alone.
The source does not fully spell out team-workspace or account-sharing behavior.
Several workflow details, such as exact setup steps and integration examples, are only partially covered in the available evidence.

FAQ

What is Featherless used for?

Featherless is a serverless AI inference platform that exposes open-weight models through API endpoints. The overview and docs point to API access for model inference, with quickstart material for making a first API call.

How do developers interact with the platform?

The docs say Featherless provides a continually expanding library of open-weight models and references common API endpoints such as `/v1/models`, `/v1/chat/completions`, `/v1/completions`, and `/v1/tokenize`.

Does Featherless have paid plans?

The pricing page shows paid plans rather than a free-only product: Basic is listed at $10 per month and Premium at $25 per month. The source does not provide a full pricing matrix beyond those plan summaries.

Are there usage limits?

The documentation includes a concurrency-limits section and the pricing page lists connection caps on the visible plans, so usage is plan-dependent. The source does not provide enough detail to describe broader account-sharing or team-workspace behavior.

Who is Featherless for?

The site positions Featherless for open-model hosting and API use, while also pointing users to Phoenix for chatting with models on the Featherless site. That suggests the platform serves both API builders and people who want to try models directly.

Quick Facts

Category: Serverless LLM hosting
Platform: Web and API
Primary users: Developers building with open-weight models
Domain: featherless.ai
Notable workflows: Model browsing, API inference, chat completions, completions, tokenization, tool calling
Pricing signal: Paid plans are listed, including Basic and Premium