OpenAI-compatible Chat Completions

Status: Stable public API.

NexoRouter's primary public integration path is the OpenAI-compatible POST /v1/chat/completions endpoint. Most SDKs and tools should start here.

Use this when

You already use the OpenAI SDK and want to change provider with minimal code changes.
A tool asks for an OpenAI-compatible, Custom OpenAI, or OpenAI API provider.
Your app sends text chat requests through messages.
You want request IDs, cost, latency, and result status to appear in Usage Logs.

Do not use this page as proof that Responses API, native Claude, native Gemini, embeddings, streaming, vision, image, video, or audio endpoints are supported.

Endpoint

Base URL: https://api.nexorouter.com/v1
Endpoint: POST /v1/chat/completions
Header: Authorization: Bearer YOUR_NEXOROUTER_API_KEY

SDK setup

Client	Key setting	Base URL setting
Python `openai`	`api_key`	`base_url`
Node.js `openai`	`apiKey`	`baseURL`
Tools using environment variables	`OPENAI_API_KEY`	`OPENAI_BASE_URL`
Generic HTTP clients	`Authorization: Bearer ...`	full request URL

For tools that reuse OpenAI environment variable names, set:

export OPENAI_API_KEY="$NEXOROUTER_API_KEY"
export OPENAI_BASE_URL="https://api.nexorouter.com/v1"

Keep those variables scoped to the shell, service, or project that should use NexoRouter.

Minimal request

curl https://api.nexorouter.com/v1/chat/completions \
  -H "Authorization: Bearer $NEXOROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      { "role": "user", "content": "Write one short production checklist." }
    ],
    "max_tokens": 256
  }'

SDK migration rule

When moving existing OpenAI SDK code, keep the request body shape and change only:

apiKey  -> NexoRouter API key
baseURL -> https://api.nexorouter.com/v1
model   -> a model ID from NexoRouter Models

Python:

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_NEXOROUTER_API_KEY",
    base_url="https://api.nexorouter.com/v1",
)

Node.js:

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.NEXOROUTER_API_KEY,
  baseURL: "https://api.nexorouter.com/v1",
});

Go:

package main

import (
    "context"
    "fmt"
    "os"

    "github.com/openai/openai-go"
    "github.com/openai/openai-go/option"
)

func main() {
    client := openai.NewClient(
        option.WithAPIKey(os.Getenv("NEXOROUTER_API_KEY")),
        option.WithBaseURL("https://api.nexorouter.com/v1"),
    )

    completion, err := client.Chat.Completions.New(
        context.Background(),
        openai.ChatCompletionNewParams{
            Model: "deepseek-v4-flash",
            Messages: []openai.ChatCompletionMessageParamUnion{
                openai.UserMessage("Write one short production checklist."),
            },
        },
    )
    if err != nil {
        panic(err)
    }

    fmt.Println(completion.Choices[0].Message.Content)
}

.NET:

using OpenAI;
using OpenAI.Chat;
using System.ClientModel;

ChatClient client = new(
    model: "deepseek-v4-flash",
    credential: new ApiKeyCredential(
        Environment.GetEnvironmentVariable("NEXOROUTER_API_KEY")
    ),
    options: new OpenAIClientOptions
    {
        Endpoint = new Uri("https://api.nexorouter.com/v1")
    }
);

ChatCompletion completion = client.CompleteChat("Write one short production checklist.");
Console.WriteLine(completion.Content[0].Text);

Other languages:

Use an OpenAI-compatible client only if it lets you set both:
api key  = NexoRouter API key
base URL = https://api.nexorouter.com/v1

Request body boundary

Start with the common Chat Completions fields:

Field	Guidance
`model`	Use an exact model ID from Models or `GET /v1/models`.
`messages`	Send OpenAI-style chat messages.
`temperature`	Optional. Keep defaults for first tests.
`max_tokens`	Optional. Use a small value for setup checks.
`stream`	Do not rely on streaming unless the relevant page documents support for your client and model.
`tools`	Verify per model and tool before production agent workflows.

Verify

After the first request, open Usage Logs and confirm the API key, model ID, result status, token counts, cost, latency, and request ID.

Common failures

Failure	Likely cause	Fix
`invalid_api_key`	Wrong key value, missing `Bearer`, expired key, disabled key, or key from another service	Check the key and the base URL together.
`model_not_found`	Model ID typo or the key's model scope excludes that model	Copy the model ID from Models and create a replacement key if scope is too narrow.
`insufficient_quota`	Account balance or key budget is not enough	Add balance or create a replacement key with the right budget.
Timeout	Slow model, short client timeout, network proxy, or large request	Raise the client timeout or choose a faster model.

Wrong vs right

Wrong	Right
NexoRouter key with `https://api.openai.com/v1`	NexoRouter key with `https://api.nexorouter.com/v1`
OpenAI key with `https://api.nexorouter.com/v1`	NexoRouter key with `https://api.nexorouter.com/v1`
Full key pasted into support chat	Key name, last four characters, and request ID