LobeChat custom OpenAI API setup: Base URL, API key, and model

Before you start

You need a RutaAPI account, a prepaid credit pack, and an API key. Make sure your account has credits before making requests — requests without sufficient credits will return a 401 or 403 error.

How to start with RutaAPI

curl https://api.rutaapi.com/v1/models \
  -H "Authorization: Bearer YOUR_RUTAAPI_KEY"

OPENAI_API_KEY=YOUR_RUTAAPI_KEY
OPENAI_PROXY_URL=https://api.rutaapi.com/v1

curl https://api.rutaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_RUTAAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_ENABLED_MODEL_ID",
    "messages": [
      {"role": "user", "content": "Hello from RutaAPI"}
    ]
  }'

Who this guide is for

This guide is for LobeChat / LobeHub users who want to connect their self-hosted chat interface to an OpenAI-compatible API gateway provided by RutaAPI.

It helps if you are familiar with environment variables, .env files, Docker or Vercel deployments.

Who should not use this setup

• Users whose LobeChat deployment does not support custom model provider configuration.
• Users whose hosting platform does not support custom environment variables.
• Users who require official OpenAI-only features.

Testing before you start chatting

Before starting a chat session in LobeChat, send a simple request from the terminal to confirm your key, proxy URL and model name all work together. This rules out configuration issues before you begin.

If the first request succeeds, the LobeChat configuration should also work. If it fails, fix the terminal request first — that gives you clearer error messages than the LobeChat UI.

How to pick a model name

Model names are returned by the GET /v1/models endpoint. Copy one of the model id values from the response and use it as the model name in your LobeChat settings.

Model availability is per-account. If a model name is not in your /v1/models response, it is not active in your account — add more credits or contact support.

Common mistakes

• Using https://rutaapi.com as OPENAI_PROXY_URL — that is the marketing site, not the API root.
• Using https://app.rutaapi.com as OPENAI_PROXY_URL — that is for account management, not the API.
• Using https://api.rutaapi.com without /v1 when LobeChat expects an OpenAI-compatible /v1 endpoint.
• Pasting https://api.rutaapi.com/v1/chat/completions as OPENAI_PROXY_URL — that is an endpoint, not the root.
• Setting a RutaAPI key but keeping another provider's proxy URL.
• Setting OPENAI_PROXY_URL correctly but using an API key from another platform.
• Guessing model names instead of using /v1/models.
• Forgetting to restart or redeploy LobeChat after changing environment variables.
• Exposing API keys in screenshots or public repositories.

Empty response troubleshooting

If LobeChat returns an empty response, the most common causes are:

• OPENAI_PROXY_URL does not include /v1 when the deployment expects it.
• The model name does not match a model ID from /v1/models.
• The account has no credits — requests may appear to succeed but return no content.
• A deployment layer or proxy is stripping or modifying the /v1 suffix.

Run /v1/models to verify the model ID, and check whether your LobeChat deployment adds /v1 automatically.

Common errors

401 Unauthorized / invalid API key

The API key is missing, incorrect, or sent to the wrong URL. Verify OPENAI_API_KEY and OPENAI_PROXY_URL both come from RutaAPI. Run /v1/models to confirm the key is valid.

403 Forbidden

The key exists but lacks permission. Check credits and model access in your RutaAPI account.

404 Model not found

The model name is not in your account. Run GET /v1/models and copy a returned model id.

429 Too Many Requests

Rate or concurrency limit. Lower the number of concurrent requests and retry.

503 Service Unavailable

The provider route is temporarily unavailable. Try a short request, check model availability, contact support if persistent.

Empty response

Usually caused by a proxy URL suffix mismatch or model name not matching /v1/models. Check whether /v1 is required and verify the model ID is in /v1/models.

When RutaAPI may be a good fit

• You need an OpenAI-compatible proxy URL for LobeChat
• You want dashboard-generated API keys
• You want prepaid credits
• You want to verify available models with /v1/models
• You need docs for LobeChat, Open WebUI, NextChat, Cursor, Cline, Continue.dev, Codex CLI and Claude Code
• You want usage visibility before scaling

When RutaAPI may not be the right fit

• Your LobeChat deployment does not load environment variables
• The selected model is unavailable upstream
• Your API key belongs to another provider
• Your deployment strips or duplicates /v1
• You require official OpenAI-only features

Related setup and troubleshooting guides

• API Key and Base URL explained
• Unified Base URL for AI coding tools
• Open WebUI OpenAI-compatible API setup
• NextChat custom API setup guide
• Cursor custom API setup guide
• Cline OpenAI-compatible API setup
• Continue.dev OpenAI-compatible API setup
• Codex CLI invalid_api_key troubleshooting
• Claude Code 524/503 timeout troubleshooting
• Available models
• Pricing and credit packs