POST /chat/completions

The Chat Completions endpoint powers conversational agents that rely on the OpenAI-compatible chat contract. Werker Health keeps the exact same request and response payloads, making migrations frictionless.

Base URL: https://api.werker.health/v1

Request Overview

Field	Type	Required	Description
model	string	✅	Any Werker-enabled chat model (e.g., gpt-4o, gpt-4o-mini, o4-mini).
messages	array	✅	Conversation history. Each item includes role (system, user, assistant, tool) and content.
temperature, top_p, presence_penalty, frequency_penalty	number	Optional	Sampling controls identical to OpenAI.
response_format	object	Optional	Return structured JSON ({"type": "json_object"}) without leaving the chat API surface.
tools, tool_choice	array / object	Optional	Register functions/actions accessible to the model.
stream	boolean	Optional	Stream tokens as they are generated over SSE.

All other OpenAI Chat Completions parameters, including logprobs, max_tokens, and stop, are supported unchanged.

Quick Example

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.werker.health/v1",
    api_key="WERKER_HEALTH_API_KEY",
)

completion = client.chat.completions.create(
    model="gpt-5",
    messages=[
        {"role": "system", "content": "You are a HIPAA-compliant triage assistant."},
        {"role": "user", "content": "My throat is sore and I have a mild fever."},
    ],
    response_format={"type": "json_object"},
)

print(completion.choices[0].message.content)

TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.werker.health/v1",
  apiKey: "WERKER_HEALTH_API_KEY",
});

const completion = await client.chat.completions.create({
  model: "gpt-5",
  messages: [
    { role: "system", content: "You are a HIPAA-compliant triage assistant." },
    { role: "user", content: "My throat is sore and I have a mild fever." },
  ],
  response_format: { type: "json_object" },
});

console.log(completion.choices[0].message.content);

Streaming Chat

Enable stream: true to receive incremental updates. Each streamed delta is protected by Werker’s two-way PHI scrubbing and persisted to immutable logs for auditability.

Tool Calls

{
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "lookup_medication",
        "description": "Return dosage guidance for the supplied medication name.",
        "parameters": {
          "type": "object",
          "properties": {
            "medication": { "type": "string" }
          },
          "required": ["medication"]
        }
      }
    }
  ],
  "tool_choice": "auto"
}

Function calling semantics and tool execution callbacks follow OpenAI’s implementation. Werker Health records each invocation and redacted payload for audit consistency.

Compliance Benefits

• Role-based access controls: Lock down who can issue chat completions containing PHI.
• Immutable audit history: Every prompt, tool call, and model response is recorded for compliance review.
• Zero retention guarantee: Werker-routed models are configured never to retain your data for training.

When to Use

• Digital triage, member support, and clinician co-pilots with conversational UX.
• Workflows that already rely on chat.completions and need a drop-in compliant backend.
• Applications where you want deterministic tool invocation without learning the Responses API yet.

For multimodal, tool-heavy workflows with unified output handling, consider the create-model-response endpoint instead.