Question 1

What is Swan Inference?

Accepted Answer

Swan Inference is a decentralized AI inference marketplace. Developers call one OpenAI-compatible API for frontier and open-source models and pay per token from a prepaid balance. Independent GPU providers serve the requests and are paid a published per-token payout.

Question 2

How do I use the Swan Inference API?

Accepted Answer

Sign up at inference.swanchain.io, verify your email, and create an API key. Point any OpenAI client at https://api.swanchain.io/v1 with that key as the bearer token. No other change is needed because the API is OpenAI-compatible.

Question 3

How much does Swan Inference cost?

Accepted Answer

Inference is priced per token, per model, at a rate the platform publishes on each model page. There is no auction and no percentage commission: consumers pay the listed price and providers are paid the listed payout. Prepaid top-ups earn bonus credit at higher tiers.

Question 4

How do I earn by providing a GPU to Swan Chain?

Accepted Answer

Run a model server such as vLLM, SGLang, llama.cpp or Ollama, then run the open-source computing-provider agent in front of it. The agent connects outbound over WebSocket, so no public IP, domain or TLS is required. You are paid a published payout price per token for the inference you serve. Any GPU with 8 GB or more works, including Apple Silicon.