Quickstart

Use this guide to send a small text request through ApiFlux and confirm that the gateway recorded it.

What you will build

You will choose an interface model ID, call it through one of the supported API protocols, and then find the request in your ApiFlux logs.

Before you start

1. Create an ApiFlux API key. Use a non-production key while evaluating the gateway.
2. Choose a model and copy an Interface model ID that supports the protocol you plan to use.
3. Open a terminal with cURL, Python 3, or Node.js 18+ available.

Set these environment variables before running an example:

export APIFLUX_BASE_URL="https://apiflux.ai"
export APIFLUX_API_KEY="YOUR_APIFLUX_API_KEY"
export APIFLUX_MODEL="YOUR_INTERFACE_MODEL_ID"

Run Python examples with python3 quickstart.py and TypeScript examples with npx tsx quickstart.ts.

Send your first request

Choose the protocol that matches your existing SDK or request format. OpenAI-compatible is the recommended starting point when you do not have a protocol requirement.

curl "${APIFLUX_BASE_URL}/v1/chat/completions" \
  --header "Authorization: Bearer ${APIFLUX_API_KEY}" \
  --header "Content-Type: application/json" \
  --data @- <<JSON
{
  "model": "${APIFLUX_MODEL}",
  "messages": [
    {
      "role": "user",
      "content": "Reply with exactly: ApiFlux is ready."
    }
  ]
}
JSON

{
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "ApiFlux is ready."
      }
    }
  ]
}

curl "${APIFLUX_BASE_URL}/v1/messages" \
  --header "x-api-key: ${APIFLUX_API_KEY}" \
  --header "anthropic-version: 2023-06-01" \
  --header "Content-Type: application/json" \
  --data @- <<JSON
{
  "model": "${APIFLUX_MODEL}",
  "max_tokens": 64,
  "messages": [
    {
      "role": "user",
      "content": "Reply with exactly: ApiFlux is ready."
    }
  ]
}
JSON

{
  "content": [
    {
      "type": "text",
      "text": "ApiFlux is ready."
    }
  ],
  "stop_reason": "end_turn"
}

curl "${APIFLUX_BASE_URL}/v1beta/models/${APIFLUX_MODEL}:generateContent" \
  --header "x-goog-api-key: ${APIFLUX_API_KEY}" \
  --header "Content-Type: application/json" \
  --data @- <<JSON
{
  "contents": [
    {
      "role": "user",
      "parts": [
        { "text": "Reply with exactly: ApiFlux is ready." }
      ]
    }
  ]
}
JSON

{
  "candidates": [
    {
      "content": {
        "role": "model",
        "parts": [
          { "text": "ApiFlux is ready." }
        ]
      }
    }
  ]
}

Verify the request

Open Logs in the ApiFlux console and find the request you just sent. Confirm the model, timestamp, token usage, and request status match the response from your terminal.

Keep the X-Request-Id response header when a request fails. It identifies the same request in ApiFlux logs and is the value support needs to investigate the failure.

Troubleshooting

Next steps