Migrate from OpenAI to TokonLab in Under 5 Minutes

If you're already using the OpenAI SDK, switching to TokonLab requires exactly one change: your base URL. That's it. No new SDK, no new abstractions, no migration guide to read for hours.

Here's the complete migration for the three most common setups.

Python

from openai import OpenAI

# Before
client = OpenAI(api_key="sk-openai-your-key")

# After — one line change
client = OpenAI(
    base_url="https://api.tokonlab.com/v1",
    api_key="sk-tokon-your-key",
)

response = client.chat.completions.create(
    model="deepseek/deepseek-r1",   # or any model alias
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

Node.js / TypeScript

import OpenAI from 'openai';

// Before
const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

// After
const client = new OpenAI({
  baseURL: 'https://api.tokonlab.com/v1',
  apiKey: process.env.TOKONLAB_API_KEY,
});

const res = await client.chat.completions.create({
  model: 'qwen/qwen3-235b-a22b',
  messages: [{ role: 'user', content: 'Hello' }],
});

cURL

curl https://api.tokonlab.com/v1/chat/completions \
  -H "Authorization: Bearer sk-tokon-your-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-r1",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Model Aliases

You can use full model IDs (like deepseek/deepseek-r1) or our convenience aliases:

• cheap-model — routes to the best value model (currently DeepSeek R1)
• fast-model — routes to the lowest latency model (currently GLM-5 Turbo)
• best-model — routes to the highest quality model (currently Qwen3 235B)

Streaming

Streaming works identically to OpenAI. Just pass stream: true and handle the response the same way you always have.

Get your API key at tokonlab.com/dashboard. Free tier included, no credit card required.