Cookbook / Quickstart

Migrating from OpenAI in 2 lines

NexToken is OpenAI-SDK compatible. You change two strings and your existing code routes to nex-pro on Singapore-hosted GPUs at ~95% lower cost.

⏱ 5 minnex-proPython · Node · curlNo new SDK

What you'll build

A chat call that hits NexToken instead of OpenAI's API — same OpenAI Python or Node SDK, no new dependency. By the end you'll have:

Step 1 · Get a key

Sign up at app.nextoken.biz. New accounts get 1,000 free nex-pro calls/day — no card required, no expiry until you top up. Your default API key is shown right after registration.

Step 2 · Change two lines

Python (OpenAI SDK ≥ 1.0)

from openai import OpenAI

client = OpenAI(
    api_key="nex_live_YOUR_KEY",                # ← change
    base_url="https://api.nextoken.biz/v1",     # ← change
)

response = client.chat.completions.create(
    model="nex-pro",                            # Singapore-hosted, $0.10/1M input
    messages=[{"role": "user", "content": "Explain quantum computing in 3 sentences."}],
)
print(response.choices[0].message.content)
print(f"Cost: ${response.nex.cost_usd}")        # Nex extension — paid in USD

Node.js / TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey:  "nex_live_YOUR_KEY",
  baseURL: "https://api.nextoken.biz/v1",
});

const r = await client.chat.completions.create({
  model: "nex-pro",
  messages: [{ role: "user", content: "Explain quantum computing in 3 sentences." }],
});
console.log(r.choices[0].message.content);

curl

curl https://api.nextoken.biz/v1/chat/completions \
  -H "Authorization: Bearer $NEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nex-pro",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Step 3 · Use any other model

NexToken supports 44 models across 10 providers through the same endpoint — GPT-4o, Claude, Gemini, DeepSeek, Qwen, GLM, and more. Just pass the model name: 

client.chat.completions.create(model="gpt-4o",        messages=[...])       # OpenAI
client.chat.completions.create(model="claude-sonnet-4-6", messages=[...])   # Anthropic
client.chat.completions.create(model="gemini-2.5-pro", messages=[...])      # Google
client.chat.completions.create(model="nex-auto",      messages=[...])       # let Nex pick
Why nex-pro by default? It's self-hosted in Singapore on our own GPU — the lowest latency in APAC, ~95% cheaper than GPT-4o, fully PDPA-compliant. For most chat / code / content tasks it's the right starting model. Switch to nex-reasoning for math/logic, or nex-auto to let the gateway choose per-request.

Step 4 · Verify the savings

Every response includes a nex.cost_usd field — the exact USD billed to your wallet. Compare it to OpenAI's per-request cost; for a 100-token round trip you should see something like: 

// nex-pro:    $0.000019
// gpt-4o:     $0.000380  (20× more)
// gpt-4o-mini: $0.000022 (still 15% more than nex-pro)

What's next