Chat Completions

Paths

Request example

curl https://dingjiai.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_USER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "A_MODEL_NAME_COPIED_FROM_THE_OPENAI_CHAT_MODEL_LIST",
    "messages": [
      { "role": "user", "content": "Introduce yourself in one sentence" }
    ]
  }'

Streaming

curl -N https://dingjiai.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_USER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "A_MODEL_NAME_COPIED_FROM_THE_OPENAI_CHAT_MODEL_LIST",
    "stream": true,
    "messages": [
      { "role": "user", "content": "Explain the integration in three points" }
    ]
  }'

The gateway prefers upstream stream usage. If at least one upstream/model chunk has been delivered but final usage is missing or invalid, a token-priced product may settle with a conservative lower-bound estimate. A disconnect or client cancellation after delivery follows the same partial-usage/output-floor rule. A truly blank or malformed non-stream response has not been committed to the client, so it fails for the client as usage_missing or usage_invalid, releases the reservation, and is charged zero.

For a stream, the gateway commits the response headers after the upstream returns 2xx. If the stream ends without delivering an upstream/model chunk, the client still receives a 2xx response; its body may be empty or contain only gateway-generated SSE keepalives. The gateway cannot rewrite that response as an error. It releases the reservation, charges zero, and records the settlement failure and call observation. This is a different client-visible result from a blank or malformed non-stream response.

Request fields

Usage and billing

Settlement reads usage.prompt_tokens, usage.completion_tokens, and the optional usage.prompt_tokens_details.cached_tokens field from the response body.

Upstream explicit usage data remains the preferred and most accurate settlement source.