02 / GET CONNECTED

Connect your workflow in three steps.

Your endpoint, key and model list arrive by private email.
Follow the setup instructions included with your delivery.

Get your dedicated key

After plan confirmation and payment, we manually deliver your key, credit and available model list.

Set your endpoint and model

Use the Base URL from your delivery email in an OpenAI-compatible client. Enter the exact model ID provided.

Test first, then start building

Send a short request and check usage and balance. Contact Run7 for manually processed top-ups.

Examples read keys from environment variables. Never put real keys in frontend code, screenshots or public repositories.

Protocol essentials

Route	What to check
Chat Completions	Basic text requests in compatible clients
Responses	Required by certain models for tool calling
Messages	Native-style integrations such as Claude Code
Gemini native	Used by Gemini CLI integrations

Protocol availability depends on your model and key. A converted protocol may handle tools, reasoning or multimodal input differently. Your private delivery guide identifies the appropriate endpoint and route.

Diagnose a failed request

Status	First action
400	Validate JSON, required fields and content type
401	Check the key and whether it was disabled
403	Check credit, model permissions and HTTPS
404	Check route and duplicated /v1 paths
413	Reduce request size
429	Reduce concurrency and respect Retry-After
502 / 503	Use bounded backoff; contact support if persistent

Retry deliberately

For 429, 502 and 503, retry with an increasing delay and a maximum attempt count. Respect Retry-After. Fix authentication and request errors before retrying. A timeout can leave completion status uncertain; do not launch unlimited duplicate requests.

Send a useful support report

Include time and timezone, tool version, model ID, status code and request ID. Remove the Authorization header, full key and private prompt content.

API REFERENCE / QUICK REQUEST

Make your first request

For macOS / Linux shells with Python 3 and cURL. Replace all placeholders with your private delivery values; the example endpoint is not a real service.

# macOS / Linux shell. Set these from your private delivery.
export RUN7_BASE_URL="https://YOUR_ENDPOINT/v1"
export RUN7_API_KEY="YOUR_PRIVATE_KEY"
export RUN7_MODEL="YOUR_MODEL_ID"

curl "${RUN7_BASE_URL%/}/chat/completions" \
  -H "Authorization: Bearer $RUN7_API_KEY" \
  -H "Content-Type: application/json" \
  -d "$(python3 -c 'import json,os; print(json.dumps({"model":os.environ["RUN7_MODEL"],"messages":[{"role":"user","content":"Hello"}]}))')"

Request and response

Use Bearer authentication with a JSON body. Set model to an exact model ID; messages contains role and content. A basic text response is read from choices[0].message.content. Inspect usage for token counts and check final charges in your customer portal. Configure tools, streaming and other protocols according to model support.