Quickstart

1. Get your API key#

Sign up at opengateway.ai, open API Keys, and click Create Key. Copy the key immediately. It is only displayed once.

export OPENGATEWAY_API_KEY="YOUR_API_KEY"

2. Make your first call#

Pick a model, pick a question, and run it right here. The code updates as you choose, and it is the same code you would paste into your own project — if you already use the OpenAI SDK, changing base_url is the only edit you make.

curl https://apis.opengateway.ai/v1/chat/completions \
  -H "Authorization: Bearer $OPENGATEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-4o-mini",
    "messages": [{"role": "user", "content": "Hello, opengateway!"}],
    "stream": true
  }'

Turn stream off and run it again. The answer is identical; what changes is that you wait for all of it before seeing any of it.

3. Swap the model#

You just did this above. The same OpenAI-compatible request shape works across every supported provider — only the model field changes:

openai/gpt-4o-mini        # cheap and fast
openai/gpt-4o             # balanced
anthropic/claude-sonnet-4 # top-tier reasoning
google/gemini-2.5-pro     # long context and multimodal

The model catalog is available from the Models API reference.

4. See what just happened#

Open the dashboard. Your request is already there, along with its token counts, cost, latency, and the provider that served it. You did not have to install anything.

Next steps#

• Streaming — Receive tokens as the model generates them.
• Fallbacks — Configure chat-completion fallback models.
• Observability — See cost, latency, and every request in one place.
• Key Features — Namespacing, parameter normalization, debug mode, and more.