FAQ

Connection Issues

How do I verify the relay is working?

Use cURL to quickly test:

curl https://hboom.ai/v1/models \
  -H "Authorization: Bearer sk-xxxxxxxxxxxx"

If you get a JSON list of models, the connection is working.

Getting 401 Unauthorized

Possible Causes

• API Key is incorrect (check for leading/trailing spaces)
• API Key has expired or been disabled
• API Key does not have permission for the requested model

Solution: Log in to the console and check the API Key status. Regenerate if necessary.

Getting 429 Too Many Requests

Note

Your request rate has exceeded the rate limit.

Solutions:

• Wait and retry
• Check your rate limit quota in the console
• Contact admin if you need a higher quota

Getting 502 / 503 Errors

Possible cause: The upstream API service is temporarily unavailable.

Solution: Wait a few minutes and retry. Contact admin if it persists.

Configuration Issues

Does the Base URL need /v1?

It depends on the tool:

Tool	Base URL Format
Cursor	https://hboom.ai/v1 ✅ needs /v1
Claude Code	https://hboom.ai ✅ no /v1 needed
Cline	https://hboom.ai/v1 ✅ needs /v1
Continue	https://hboom.ai/v1 ✅ needs /v1
Aider	https://hboom.ai/v1 ✅ needs /v1
ChatBox	https://hboom.ai ✅ no /v1 needed
Cherry Studio	https://hboom.ai ✅ no /v1 needed

Rule of thumb

Most tools need the /v1 suffix. Anthropic-based tools and some desktop clients append it automatically.

Environment variables and in-app settings conflict — what do I do?

In-app settings typically take higher priority. Use only one method to avoid confusion. If unsure, clear the environment variables and use in-app settings only.

Which models are supported?

Visit the console or call the /v1/models endpoint for the full list. Common models include:

Model	Description
gpt-4o	OpenAI GPT-4o, multimodal
gpt-4.1	OpenAI GPT-4.1, latest
gpt-4o-mini	Lightweight GPT-4o for autocomplete
claude-sonnet-4-20250514	Claude 4 Sonnet, strong at coding
claude-opus-4-20250514	Claude 4 Opus, best reasoning
gemini-2.5-pro	Google Gemini 2.5 Pro