Connecting apps

Any tool that speaks the OpenAI-compatible API can be pointed at RelayModels by swapping the base URL and key. Below are ready-made settings for popular tools.

Automatic setup

On the API keys page in your dashboard there's a “Connect app” button. Pick the tool and your OS — the installer writes the URL and key into the right config files and keeps a backup so you can roll it back.

Buyers who redeemed a code get the same option in their dashboard, on the “Keys” and “Documentation” pages.

Manual setup

Codex CLI and other terminal agents

Configured as a compatible provider: base URL and key.

base_url = "https://api.relaymodels.com/v1"
env_key  = "RELAY_API_KEY"
model    = "gpt-5.6-sol"

Claude Code

Passed in via environment variables before launch.

export ANTHROPIC_BASE_URL=https://api.relaymodels.com/v1
export ANTHROPIC_AUTH_TOKEN=$RELAY_API_KEY
export ANTHROPIC_MODEL=claude-sonnet-4-6

Cursor, Continue, Zed, and any OpenAI-compatible client

Pick “custom provider” / Custom base URL in settings.

Base URL: https://api.relaymodels.com/v1
API Key:  <your RelayModels key>
Model:    gpt-5.6-sol

n8n, Make, LangChain

Use the OpenAI node and swap the base URL.

Node type: OpenAI (Custom base URL)
Base URL: https://api.relaymodels.com/v1
Model:    gemini-3.7-flash

Telegram bot: voice transcription

The file is sent as a regular multipart form.

curl https://api.relaymodels.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $RELAY_API_KEY" \
  -F model=gpt-4o-transcribe \
  -F file=@voice.ogg

Checking the connection

If a tool doesn't respond, start with a direct request — it will immediately show what's wrong.

curl https://api.relaymodels.com/v1/models -H "Authorization: Bearer $RELAY_API_KEY"

Got a list of models — the key and URL are correct, the issue is in the app's settings. Got an error — see Errors & limits.

What's next

Models & picking one →

What to put in the model field.

Keys & access →

Separate keys and limits.