Migration Guide

Yunxin speaks the wire formats you already use. For most codebases, migrating is a two-line change: point your SDK at https://api.yuhuanstudio.com/v1 and swap in a Yunxin API key. Your existing request/response handling, streaming loops, and tool-calling code keep working.

Before you start

• Get a key. Create one in the Dashboard → API Keys. Yunxin keys begin with sk- (see Authentication).
• Pick the closest format. Yunxin natively supports three request formats — keep using the one that matches your current code so the migration stays a config-only change:

• Note the base URL difference between SDKs. The OpenAI SDK expects the /v1 suffix; the Anthropic SDK appends /v1/messages itself, so its base URL has no /v1:

From OpenAI

Change the base_url and api_key. Everything else — Chat Completions, streaming, function calling, vision, embeddings, image/audio generation, files, and batches — works unchanged.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["YUNXIN_API_KEY"],        # an sk-... key
    base_url="https://api.yuhuanstudio.com/v1",  # was https://api.openai.com/v1
)

response = client.chat.completions.create(
    model="model-id",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.YUNXIN_API_KEY,            // an sk-... key
  baseURL: "https://api.yuhuanstudio.com/v1",    // was https://api.openai.com/v1
});

const response = await client.chat.completions.create({
  model: "model-id",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(response.choices[0].message.content);

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

Switching via environment variables

If your code reads the standard OpenAI environment variables, you can migrate without touching the source at all:

# Before (direct OpenAI)
export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://api.openai.com/v1"

# After (Yunxin) — still an sk-... key, just a different base URL
export OPENAI_API_KEY="sk-your-yunxin-key"
export OPENAI_BASE_URL="https://api.yuhuanstudio.com/v1"

Responses API

If you already use OpenAI's Responses API, keep using it — Yunxin implements POST /v1/responses natively, including previous_response_id for server-side multi-turn and the built-in tool list at GET /v1/responses/tools. See Responses API.

From Anthropic

Keep the Anthropic SDK and the Messages format. Set the base URL to https://api.yuhuanstudio.com without /v1 — the SDK appends /v1/messages for you.

import os
from anthropic import Anthropic

client = Anthropic(
    api_key=os.environ["YUNXIN_API_KEY"],       # an sk-... key
    base_url="https://api.yuhuanstudio.com",    # no /v1 — the SDK adds /v1/messages
)

message = client.messages.create(
    model="model-id",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}],
)
print(message.content[0].text)

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.YUNXIN_API_KEY,           // an sk-... key
  baseURL: "https://api.yuhuanstudio.com",      // no /v1 — the SDK adds /v1/messages
});

const message = await client.messages.create({
  model: "model-id",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(message.content[0].text);

Extended thinking, vision/PDF input, tool use, and prompt caching all pass through. The optional anthropic-version header is accepted. See Messages API.

From OpenRouter

OpenRouter users are already accustomed to provider/model ids and a single key fronting many providers — both carry over directly. Point the OpenAI SDK at https://api.yuhuanstudio.com/v1 with your sk- key:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["YUNXIN_API_KEY"],
    base_url="https://api.yuhuanstudio.com/v1",  # was https://openrouter.ai/api/v1
)

response = client.chat.completions.create(
    model="provider/model",  # same provider/model convention you already use
    messages=[{"role": "user", "content": "Hello!"}],
)

As a migration nicety, Yunxin also exposes OpenRouter-shaped usage endpoints (GET /v1/credits, GET /v1/key, GET /v1/activity) so dashboards and CLIs you already point at OpenRouter keep working unchanged. See Usage Monitoring Tools.

From other OpenAI-compatible providers

Any provider that exposes an OpenAI-compatible Chat Completions API migrates the same way: change api_key and base_url, and keep your model-id (use the provider/model form to pin a provider).

What you gain

After migrating

• Billing switches to Yunxin's credit system: 1 credit = $0.001 USD, drawn down per request based on the model's per-token prices. A 402 insufficient_balance error means you need to top up.
• Review Error Handling — the error envelope uses error.code, which differs slightly from each upstream provider's shape.
• Tune retries to honor Retry-After on 429 responses; see Rate Limits.