OpenAI-compatible API

Point the OpenAI SDK at Inference Cloud, or call the HTTP endpoints directly. Chat completions, streaming, tools, and model listing are available.

Endpoints

POSThttps://inferenceapi.wethinkdigital.solutions/v1/chat/completions

Create a chat completion. Supports streaming with server-sent events.

GEThttps://inferenceapi.wethinkdigital.solutions/v1/models

List the active model catalog available to your account.

GEThttps://inferenceapi.wethinkdigital.solutions/v1/models/{model}

Retrieve a single model, including limits and capabilities.

Use the base URL override

For the OpenAI SDK, set baseURL to https://inferenceapi.wethinkdigital.solutions/v1. The SDK appends /chat/completions automatically.

Chat completions

The request body follows the OpenAI chat completions schema. model must be a platform model ID. Retrieve valid IDs from GET /v1/models or the Models page in the console.

chat-completion

curl https://inferenceapi.wethinkdigital.solutions/v1/chat/completions \
  -H "Authorization: Bearer sk_live_xxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "model-pro",
    "messages": [
      {
        "role": "system",
        "content": "You are a concise technical assistant."
      },
      {
        "role": "user",
        "content": "Summarize the CAP theorem in two sentences."
      }
    ],
    "max_tokens": 256,
    "temperature": 0.4
  }'

Core parameters

Field	Type	Description
model	string	Required. A platform model ID from the active catalog.
messages	array	Required. Ordered conversation turns with role and content. System, user, and assistant roles are supported.
max_tokens	integer	Optional. Upper bound on generated output tokens. Defaults and limits depend on the model.
temperature	number	Optional. Sampling temperature. Supported range is model-dependent.
stream	boolean	Optional. When true, the response is an SSE stream of chat.completion.chunk objects.
tools	array	Optional. Tool definitions. Only available when the model capability reports tools support.

Streaming

Set stream: true to receive server-sent events. Each event contains a chat.completion.chunk with incremental content in choices[0].delta. The stream terminates with data: [DONE].

streaming

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.INFERENCE_API_KEY,
  baseURL: "https://inferenceapi.wethinkdigital.solutions/v1",
});

const stream = await client.chat.completions.create({
  model: "model-pro",
  messages: [{ role: "user", content: "Write a short release note." }],
  stream: true,
});

for await (const chunk of stream) {
  const delta = chunk.choices[0]?.delta?.content;
  if (delta) process.stdout.write(delta);
}

sse-stream

data: {"id":"chatcmpl_…","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant"}}]}

data: {"id":"chatcmpl_…","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"Hello"}}]}

data: {"id":"chatcmpl_…","object":"chat.completion.chunk","choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":{"prompt_tokens":24,"completion_tokens":8,"total_tokens":32}}

data: [DONE]

List models

Model metadata includes the platform ID, context window, max output tokens, capabilities, and pricing. Use this endpoint as the source of truth for model IDs.

list-models

curl https://inferenceapi.wethinkdigital.solutions/v1/models \
  -H "Authorization: Bearer sk_live_xxxxxxxx"

models-response.json

{
  "object": "list",
  "data": [
    {
      "id": "model-pro",
      "object": "model",
      "owned_by": "platform",
      "context_window": 200000,
      "max_output_tokens": 8192,
      "capabilities": {
        "streaming": true,
        "tools": true,
        "vision": true,
        "json_mode": true
      }
    }
  ]
}

Response fields

A successful non-streaming completion returns a standard chat.completion object. Usage is recorded against your balance per request.

completion-response.json

{
  "id": "chatcmpl_01J...",
  "object": "chat.completion",
  "created": 1770000000,
  "model": "model-pro",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The CAP theorem states that a distributed data store can..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 42,
    "completion_tokens": 118,
    "total_tokens": 160
  }
}

Field	Type	Description
id	string	Unique completion identifier.
model	string	The platform model ID that served the request.
choices[].message.content	string	Generated assistant content. May be empty when a tool call is returned.
choices[].finish_reason	string	Why generation stopped: stop, length, tool_calls, or content_filter.
usage	object	prompt_tokens, completion_tokens, and total_tokens used for billing.

Tools & structured output

Tool calling and JSON mode are enabled per model. Check the capabilities object returned by GET /v1/models before depending on them.

capability-check

{
  "capabilities": {
    "streaming": true,
    "tools": true,
    "vision": false,
    "json_mode": true
  }
}

Requests using an unsupported capability return a structured 400 error. Model IDs and capability flags can change when the catalog is updated; resolve them at startup or cache them with a short TTL.

All documentation Create an API token →