Quick Start

Two ways to integrate APIYI — hand the docs to your AI coding agent and let it do the work, or wire it up yourself in three steps.

New accounts come with $0.05 in trial credit, so you can make the first call below without topping up. A lightweight model such as gpt-5.4-mini is more than enough to confirm your integration works.

Two paths, pick one

Let an AI do it

Copy one prompt into Codex, Claude Code or Cursor. It reads the docs, writes the code and runs it. Best if you already work with a coding agent.

Do it yourself

Register, create a key, make your first call. Three steps, five minutes. Best if you want to understand each piece.

Both paths start the same way: you register the account and create the key yourself. What an AI can take over is everything after that — picking a model, getting the base URL right, writing the example, and debugging it.

Let an AI do it

Send your agent this prompt

Let a coding agent integrate APIYI on its own. Copy and paste into Codex, Claude Code, Cursor and similar tools.

CopiedCopy prompt

What it will do

1. Install the skill, or read skill.md directly

   https://docs.apiyi.com/skill.md is an integration reference written for machines: endpoints, authentication, the model-naming rule, known pitfalls and a verification checklist. Reading it gives an agent everything it needs to integrate APIYI.

2. Look up pages as needed

   llms.txt is the index of every page on this site. The agent picks the relevant ones and appends .md to read them as plain text.

3. Write code and actually run it

   It writes the example in your project’s existing language and dependencies rather than copying the Python from the docs. It is not done until the call succeeds.

Five entry points you can feed to an AI

Entry point	URL	When to use it
Skill	https://docs.apiyi.com/skill.md	Give an agent the complete integration reference in one shot. Start here.
Page index	https://docs.apiyi.com/llms.txt	Let the agent decide which page it needs
Single page as text	Append .md to any documentation URL	When you only care about one page — cheaper than the HTML
MCP server	https://docs.apiyi.com/mcp	Connect this documentation site as an MCP server so your agent can search it live
Model registry	https://docs.apiyi.com/model-registry.json	Machine-readable model IDs, endpoints, groups and list prices, so the agent stops guessing

The skill, the CLI and the contract-plus-registry path for coding agents are all described in the AI Developer Kit.

The plain-text version of this page, for example, is https://docs.apiyi.com/en/getting-started.md. Every page on the site supports the suffix.

Send any page straight to an AI

Every documentation page has a Copy page button in the top right. Opening the arrow next to it reveals more options:

Option	What it does
Copy page	Copies the current page as Markdown to your clipboard, ready to paste into any AI
View as Markdown	Opens the plain-text version in the browser — handy for checking or sharing a link
Open in ChatGPT	Jumps to ChatGPT carrying this page as context
Open in Claude	Same, for Claude
Open in Perplexity	Same, for Perplexity
Open in Google AI Studio	Same, for Google AI Studio

When you hit a problem with a specific model, the fastest route is to open that model’s page, hit Copy page, and send it to an AI along with your error.

Do it yourself

Step 1: Register and get a key

1. Create an account

   Go to the APIYI website, register with email and verify it (a university or corporate address is recommended), then sign in to the console.

2. Create an API key

   Open the token page:

   1. You can copy the default token and use it directly (copy icon on the right)
   2. Or click New in the top right to create one, name it (for example test-key), and confirm

   Keys start with sk-. See How to create a key for details.

3. Top up when you need more

   Once the $0.05 trial credit runs out, top up from the console. Minimum amounts and settlement rules differ by payment channel — see Payment Methods — and bonus policy is covered in Recharge Promotions.

Step 2: Get the connection details right

Choose the base URL by SDK, not by model. This is the most common integration mistake:

Your SDK	Base URL	Why
OpenAI SDK (and most clients)	https://api.apiyi.com/v1	The SDK appends /chat/completions itself, so /v1 must be there
Anthropic SDK (Claude native)	https://api.apiyi.com	The SDK appends /v1/messages itself — adding /v1 produces /v1/v1/messages and a 404
Google GenAI SDK (Gemini native)	https://api.apiyi.com	Also set api_version: "v1beta"

Do not leave a trailing slash on base_url — it produces doubled slashes and a 404. Full details and node selection are in Base URL configuration.

Step 3: Make your first call

curl https://api.apiyi.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $APIYI_API_KEY" \
  -d '{
    "model": "gpt-5.4-mini",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["APIYI_API_KEY"],
    base_url="https://api.apiyi.com/v1"
)

response = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[
        {"role": "user", "content": "Hello!"}
    ]
)

print(response.choices[0].message.content)

Node.js

import OpenAI from 'openai';

const openai = new OpenAI({
  apiKey: process.env.APIYI_API_KEY,
  baseURL: 'https://api.apiyi.com/v1'
});

const response = await openai.chat.completions.create({
  model: 'gpt-5.4-mini',
  messages: [{ role: 'user', content: 'Hello!' }]
});

console.log(response.choices[0].message.content);

Java

// Using the official OpenAI Java library
OpenAiService service = new OpenAiService(
    System.getenv("APIYI_API_KEY"),
    Duration.ofSeconds(60),
    "https://api.apiyi.com/v1"
);

ChatCompletionRequest request = ChatCompletionRequest.builder()
    .model("gpt-5.4-mini")
    .messages(List.of(
        new ChatMessage(ChatMessageRole.USER, "Hello!")
    ))
    .build();

ChatCompletionResult result = service.createChatCompletion(request);
System.out.println(result.getChoices().get(0).getMessage().getContent());

The gpt-5 series and above have three parameter restrictions: temperature must be 1, use max_completion_tokens instead of max_tokens, and do not send top_p.

Next steps

Connect Claude Code

Set ANTHROPIC_BASE_URL and drive Claude Code through APIYI

Connect Codex

One config.toml covers the desktop app, the IDE plugin and the CLI

Explore the model list

Every supported model and a capability cheat sheet

Read the API manual

Full endpoint reference, error codes and debugging

FAQ

How do I switch models?

Just change the model parameter in your request:

{
  "model": "gpt-5.6-sol",         // Use GPT-5.6 Sol
  "model": "claude-opus-5",       // Use Claude Opus 5
  "model": "gemini-3.6-flash"     // Use Gemini 3.6 Flash
}

Model IDs use dots, not hyphens. The hyphens in documentation URLs are a URL-safety substitution; the real model ID keeps its dots. The page /models/qwen3-7-max corresponds to the model ID qwen3.7-max. Writing gpt-5-4-mini returns a 404 — the correct form is gpt-5.4-mini.

When in doubt, list them: GET https://api.apiyi.com/v1/models.

Which programming languages are supported?

APIYI is compatible with OpenAI API standards and supports all languages that OpenAI SDKs support:

• Python
• JavaScript/TypeScript
• Java
• C#/.NET
• Go
• Ruby
• PHP
• And more…

How do I check my balance?

Sign in to the console to view:

• Account balance
• Usage history
• Consumption statistics

You can also query programmatically via API:

• Balance Query API: Get account balance, expiry date, and more via API
• Balance Alert Setup: Auto-notify on low balance to avoid service interruption

What if I run into a problem?

1. Open the page that is giving you trouble, hit Copy page in the top right, and send it to an AI together with your error
2. Check the API manual
3. Review common errors
4. Contact support: [e-mail]

Tip: Save your API key securely and regularly check usage logs in the console. Every request has message history for cost optimization. Happy using!