Setup & install

Automatic setup

Paste your key and run one command. The script detects macOS, Linux, or Windows, finds the coding tools you have installed, and points them at Nixo with your key and the right base URL. Every changed file is backed up first.

$ curl -fsSL https://nixo.codes/setup=sk-nx-YOUR_KEY | bash

Tools that are not installed are skipped. NIXO_API_KEY is also saved to your shell profile (or as a Windows user environment variable). Prefer not to put the key in the URL? Run curl -fsSL https://nixo.codes/setup | bash or irm https://nixo.codes/setup | iex and the script asks for it.

Claude Code

Anthropic-compatible clients use the Nixo root URL without /v1. Add these values to your Claude Code settings.

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://nixo.codes",
    "ANTHROPIC_AUTH_TOKEN": "your-key-here"
  }
}

Conductor can use the same Claude Code settings. You can also set ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY directly in its app environment.

export ANTHROPIC_BASE_URL=https://nixo.codes
export ANTHROPIC_API_KEY=your-key-here

Codex CLI / Desktop

Create or edit ~/.codex/config.toml and register Nixo as a Responses-compatible provider.

model = "openai/gpt-6-sol"
review_model = "openai/gpt-6-sol"
model_provider = "nixo"

[model_providers.nixo]
name = "Nixo"
base_url = "https://nixo.codes/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"

Set your key in the environment before launching Codex:

export OPENAI_API_KEY=your-key-here

To use another supported GPT model, update both top-level model fields to its ID from the model catalog. In Codex Desktop, set Base URL to https://nixo.codes/v1 and enter your Nixo API key.

SDK examples

Use the official SDKs with a custom base URL. Install the Anthropic or OpenAI SDK in your project before importing it. Streaming is supported and metered (stream: true, messages.stream()): events are relayed as they arrive and billed from the final token usage the stream reports. If a stream is cut off before it reports usage, the prompt plus the output received so far are billed as an estimate.

Python · Anthropic SDK

import anthropic

client = anthropic.Anthropic(
    api_key="your-key-here",
    base_url="https://nixo.codes",
)

message = client.messages.create(
    model="anthropic/opus-5-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(message.content[0].text)

Python · OpenAI SDK

from openai import OpenAI

client = OpenAI(
    api_key="your-key-here",
    base_url="https://nixo.codes/v1",
)

response = client.chat.completions.create(
    model="anthropic/opus-5-5",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

TypeScript · Anthropic SDK

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

const client = new Anthropic({
  apiKey: "your-key-here",
  baseURL: "https://nixo.codes",
});

const message = await client.messages.create({
  model: "anthropic/opus-5-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(message.content[0].text);

Environment variables

Anthropic-compatible clients use https://nixo.codes with no /v1. OpenAI-compatible clients use https://nixo.codes/v1.