For Developers

Step-by-step guide to using Swan 2.0 Inference Cloud as a consumer

This guide walks through using Swan Inference as a developer consuming AI models — starting with a no-signup trial, then creating an account, topping up, and making real API requests.

Looking to earn by providing GPU resources instead? See Become a Provider for the step-by-step setup guide.

0. Try it now — no signup

The fastest way to see Swan Inference in action: open the playground, pick a model, and send a message. No account, no API key, no credit card.

Playground — runs GLM 4.7 Flash for anonymous visitors, rate-limited per IP.

Ready for more? Sign up below to get an API key and start integrating.

1. Sign up and get your API key

Create a free account at inference.swanchain.io/signup — email and password only, no credit card required.

Verify your email before continuing. Signing up sends a verification link, and the account cannot sign in or issue API keys until you click it. If you would rather skip this, sign in with a wallet instead: a wallet-bound account authenticates by signature, so it needs no email verification.

Sign up with email and password.

Once verified, navigate to Keys in the dashboard. Your API key (sk-swan-*) is generated automatically — copy it and keep it secret.

Your API key appears under Keys in the dashboard.

2. Top up credits

Inference requests are paid per token, deducted from your account balance in real time. Fund your account via Stripe (credit card) or crypto deposit (USDC / USDT / SWAN on multiple EVM chains).

• Stripe: instant processing, minimum deposit $5

• Crypto: per-user HD-derived deposit address shared across EVM chains, minimum $1

Add funds via Stripe card payment or crypto deposit.

Deposit bonuses

Larger deposits are credited with extra: $100 or more → +2%, $500 or more → +4%, $2,000 or more → +6%. Depositing SWAN on Swan Mainnet earns a flat 10% instead — the two do not stack, a SWAN deposit takes the SWAN bonus.

Your account balance is a single USD-denominated pool regardless of how it was funded, so there is nothing to toggle at request time; you simply get more credits per dollar. The live schedule is GET /api/v1/subscription/deposit-bonus-tiers, and the pricing page shows the same figures — flip its Pay with: SWAN toggle to see the effective per-token rate across every model.

Flip the Pay-with toggle to SWAN to preview the effective rate after the SWAN deposit bonus.

Usage is deducted from your balance per request. View balance, usage, and the transaction ledger under Billing in the dashboard.

There is no subscription plan. Every request is pay-as-you-go at the model's list price. The monthly Token Plan (Pro) has been retired and is no longer sold; subscriptions bought before then run to the end of the period already paid for and are not renewed.

3. Browse models

The Models page lists every model with live pricing, context length, tier, and how many providers are online for it right now — a model is only callable while at least one provider is serving it. Click any model for details and code examples.

Live models catalog.

The Pricing page compares SwanChain's rates side-by-side against Anthropic, Google, OpenRouter, and other providers for hero models — so you can see how prices stack up at a glance.

Pricing page with competitor comparison.

4. Make your first inference request

Swan Inference is fully OpenAI-compatible — any existing OpenAI SDK or integration works by changing two things: the base URL and the API key. The examples below use zai-org/GLM-4.7-Flash, one of the cheapest hero models at $0.05 input / $0.36 output per 1M tokens.

curl

curl https://api.swanchain.io/v1/chat/completions \
  -H "Authorization: Bearer sk-swan-YOUR-KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zai-org/GLM-4.7-Flash",
    "messages": [{"role": "user", "content": "Say hello in 5 words."}]
  }'

OpenAI Python SDK

from openai import OpenAI

client = OpenAI(
    base_url="https://api.swanchain.io/v1",
    api_key="sk-swan-YOUR-KEY",
)

response = client.chat.completions.create(
    model="zai-org/GLM-4.7-Flash",
    messages=[{"role": "user", "content": "Say hello in 5 words."}],
)
print(response.choices[0].message.content)

OpenAI Node.js SDK

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.swanchain.io/v1",
  apiKey: "sk-swan-YOUR-KEY",
});

const response = await client.chat.completions.create({
  model: "zai-org/GLM-4.7-Flash",
  messages: [{ role: "user", content: "Say hello in 5 words." }],
});
console.log(response.choices[0].message.content);

Streaming, embeddings, image generation, and audio transcription all work identically to OpenAI. See the API reference for every endpoint, the response headers that tell you which provider served you, and the limits.

5. Optional: choose the provider

Every model page lists the providers currently serving it, with each one's price, context window, quantization, 30-day uptime and typical time-to-first-token. To send a request to a specific one, add X-Swan-Provider: <provider-id>; add X-Swan-Allow-Fallbacks: false if you would rather the request fail than be served by someone else. The response tells you who served it and how it was billed (X-Swan-Route-Mode, X-Swan-Fallback-Reason, X-Swan-Billing-Type).

Pinning changes who serves the request, not how it is billed: it is charged from your credit balance at the model's catalog price, as every request is, including when your provider is offline and a fallback serves it. Details in the API reference.

Next steps

• Inference Marketplace — deeper on how pricing, routing, and settlement work

• Become a Provider — want to earn by sharing GPU resources instead?

• API reference — full list of endpoints, headers, limits and error codes

• AI agent integrations — point OpenClaw, Nanobot and friends at Swan Inference

Questions? Reach the team on Discord or open an issue on GitHub.