Skip to main content

Chat Completion

The core endpoint of the WiLine Edge Cloud Inference API. Send a conversation (a list of messages) and receive a model-generated reply. The API is fully OpenAI-compatible — any SDK or tool that supports OpenAI works unchanged by pointing base_url at https://inference.wiline.com/v1.

POST 
/v1/chat/completions

Create a chat completion on any model available on your WiLine Edge Cloud endpoint.

Normalizer Mode: Append -normalize to the model name (e.g. deepseek-ai/DeepSeek-V4-Pro-normalize) to route the request through the normalizer proxy, which applies the following fixes to non-standard OpenAI responses:

  • reasoning_content → reasoning (streaming deltas and non-streaming messages)
  • Tool call IDs: functions.xxx:N → chatcmpl-tool-<hash>
  • content=None with tool_calls → content=""
  • Strip empty tool_calls=[] from streaming deltas
  • Remove content=null from streaming tool call deltas
  • Strip whitespace-only content when tool_calls are present in streaming deltas

OpenAI-compatible: Any OpenAI SDK or tool works unchanged — set base_url to https://inference.wiline.com/v1 and use your WEC API key as the Bearer token.

curl -X POST https://inference.wiline.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $WEC_API_KEY" \
  -d '{
    "model": "deepseek-ai/DeepSeek-V3.1",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Request​

Responses​

Successful Response