Lightweight AI Gateway

Cursor

Cursor allows overriding the default OpenAI Base URL for custom proxy routing.

1. Open Cursor Settings.
2. Navigate to Models > OpenAI.
3. Toggle "Override OpenAI Base URL".
4. Set the URL to https://api.lightweight.one/v1.
5. Paste your lw-... key in the OpenAI API Key field.
6. Ensure external infrastructure integration toggles are off.

Aider

Aider is an AI pair programming tool in your terminal. You can pass the API settings as CLI arguments or environment variables.

aider --openai-api-key "lw-your-key-here" --openai-api-base "https://api.lightweight.one/v1" --model "gpt-4o"

Alternatively, export OPENAI_API_KEY and OPENAI_API_BASE in your shell profile.

Zed

Zed's built-in AI assistant can be configured via your settings.json.

1. Open Zed and edit your settings.json.
2. Add or update the language_models block:

"language_models": {
  "openai": {
    "api_url": "https://api.lightweight.one/v1",
    "api_key": "lw-your-key-here"
  }
}

Continue

Continue is an open-source IDE extension.

1. Open your ~/.continue/config.json file.
2. In the models array, configure an OpenAI provider:

{
  "models": [
    {
      "title": "Lightweight Opus",
      "provider": "openai",
      "model": "gpt-4.1",
      "apiKey": "lw-your-key-here",
      "apiBase": "https://api.lightweight.one/v1"
    }
  ]
}

Cline

Cline perfectly matches any OpenAI-compatible endpoint.

1. Open Cline's settings panel.
2. Under API Provider, select OpenAI Compatible.
3. Set Base URL to https://api.lightweight.one/v1.
4. Paste your lw-... key in the API Key field.
5. Specify the Model ID (e.g., gpt-5).

Windsurf

Windsurf can be configured to use external API gateways.

1. Open Windsurf settings.
2. Navigate to the API or Models configuration section.
3. Select Custom API Endpoint.
4. Enter the Base URL: https://api.lightweight.one/v1.
5. Input your Lightweight API key.

Ollama

If you use tools that expect an Ollama-style proxy but want to route them through Lightweight, set these environment variables before starting the tool.

export OLLAMA_OPENAI_BASE_URL="https://api.lightweight.one/v1"
export OLLAMA_OPENAI_API_KEY="lw-your-key-here"

OpenAI SDK (Python)

The official Python SDK requires no modifications other than instantiating the client with our base URL and key.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.lightweight.one/v1",
    api_key="lw-your-key-here"
)

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

OpenAI SDK (Node.js)

Similarly, the Node.js/TypeScript SDK natively supports custom base URLs.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.lightweight.one/v1",
  apiKey: "lw-your-key-here",
});

async function main() {
  const completion = await client.chat.completions.create({
    model: "gpt-4o",
    messages: [{ role: "user", content: "Hello!" }],
  });
  console.log(completion.choices[0].message.content);
}
main();

LangChain

Use the standard ChatOpenAI class in LangChain, overriding the base URL parameter.

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.lightweight.one/v1",
    api_key="lw-your-key-here",
    model="gpt-4.1"
)

print(llm.invoke("Explain quantum entanglement.").content)

LiteLLM

When using LiteLLM to proxy requests locally, direct the OpenAI proxy target to Lightweight.

litellm --model openai/gpt-4.1 \
  --api_base "https://api.lightweight.one/v1" \
  --api_key "lw-your-key-here"

Any OpenAI-Compatible Tool

For any other application that doesn't have an explicit UI setting, environment variables are universally respected by the underlying OpenAI SDKs.

1. Find where the tool loads its environment variables (e.g., a .env file, a shell script, or system environment settings).
2. Set OPENAI_API_KEY=lw-your-key
3. Set OPENAI_BASE_URL=https://api.lightweight.one/v1
4. Restart the application.