How to Use evolink Smart router: 5-Minute Setup for Unified AI Model Routing

What evolink Smart router is

Use it when you want to:

• keep one OpenAI-compatible request format
• reduce application-side switching between providers or model families
• inspect the routed model in the API response instead of hard-coding one model everywhere
• start with a flexible gateway path before pinning a fixed model for production-critical flows

If you already know the exact model, latency profile, and cost target you need, a fixed model ID is usually the cleaner choice.

What you need before you start

Your first request with curl

curl https://api.evolink.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "evolink/auto",
    "messages": [
      {
        "role": "user",
        "content": "Explain vector databases in one short paragraph."
      }
    ]
  }'

Python example

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.evolink.ai/v1"
)

response = client.chat.completions.create(
    model="evolink/auto",
    messages=[
        {
            "role": "user",
            "content": "Summarize the tradeoff between latency and model quality."
        }
    ]
)

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

Node.js example

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.EVOLINK_API_KEY,
  baseURL: "https://api.evolink.ai/v1",
});

const response = await client.chat.completions.create({
  model: "evolink/auto",
  messages: [
    {
      role: "user",
      content: "List three reasons teams use an AI gateway.",
    },
  ],
});

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

How to read the response

A smart-router response still follows the familiar chat completions structure:

{
  "id": "chatcmpl-example",
  "object": "chat.completion",
  "created": 1773187200,
  "model": "provider/model-selected-at-runtime",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "A vector database stores embeddings so semantic search can retrieve similar content efficiently."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 17,
    "completion_tokens": 18,
    "total_tokens": 35
  }
}

Smart router vs fixed model

The pattern that usually scales best is simple:

1. Start with evolink Smart router while the workload is still changing.
2. Log the routed model value and compare cost, latency, and output quality.
3. Pin a fixed model for flows that need tighter operational predictability.

What remains account-specific

The original draft included several product claims that should not be published as hard facts without a verified source tied to your account or official documentation. Treat these as items to confirm before you publish external promises:

• the exact public model identifier for evolink Smart router
• the exact size of the available routing pool
• any promise about "no routing fee"
• any percentage discount claim versus direct vendors
• SLA statements such as 99.9% uptime
• which advanced capabilities are guaranteed through the smart router for every routed model

If you want this article to become more sales-forward later, the clean way is to add those items only after they are documented in a first-party pricing page, product page, or official API doc.

Production checklist before rollout

Next steps

Once your first request works, the highest-value follow-up is not more sample code. It is instrumentation.

• log response.model
• store token usage by feature or route
• compare smart-router traffic against one pinned-model baseline
• review available fixed models in the EvoLink model catalog

That gives you the data needed to decide whether the gateway path is improving cost efficiency and production reliability for your actual workload.

FAQ

Is evolink Smart router the same as choosing a fixed model?

Do I need a different SDK to use evolink Smart router?

No. Based on the repository's existing examples, the integration pattern stays OpenAI-compatible. You mainly change the base URL and model identifier.

Where do I find the correct Smart router model ID?

Confirm it in your EvoLink dashboard or official docs before publishing or shipping copy-paste code. The original draft did not include a locally verified identifier.

Should I start production traffic on evolink Smart router?

Yes, if your workload is still evolving and you want one gateway entry point. For tightly controlled flows, compare it against a pinned model before full rollout.

What should I log first after integration?

Does smart routing guarantee lower cost?

Not automatically. It can improve effective cost, but the result depends on your prompts, selected downstream models, and account configuration.

When should I switch from Smart router to a fixed model?

Switch when one workload has a clear winner on quality, latency, or cost and you want tighter QA and more predictable production behavior.