Responses API

Responses is OpenAI's newer API format: it uses input instead of messages and returns results in output. AI Nexus offers POST /v1/responses, so the OpenAI SDK's client.responses.create() works directly. Codex CLI also connects through this endpoint.

Basic callCopy Anchor Link

Replace your-model-id in the samples with a model ID from the model square (or call GET /v1/models).

# Responses format: path /v1/responses, input instead of messages
curl -sS https://ai.myaiexp.com/v1/responses \
  -H "Authorization: Bearer $MYAIEXP_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "your-model-id",
    "input": "Introduce yourself in one sentence.",
    "max_output_tokens": 600
  }'

Key points:

• The base URL is also https://ai.myaiexp.com/v1, with Authorization: Bearer <Key>.
• max_output_tokens corresponds to max_tokens in Chat Completions. Reasoning models count their thinking as output, so leave enough room.
• input can be a string or an array of messages similar to messages; pass a system prompt with instructions.

Response structureCopy Anchor Link

The answer is in output[].content[].text (with the SDK, read resp.output_text), and usage is in usage.input_tokens / usage.output_tokens. Structure:

{
  "id": "resp_...",
  "object": "response",
  "model": "your-model-id",
  "status": "completed",
  "output": [
    {
      "type": "message",
      "role": "assistant",
      "status": "completed",
      "content": [
        { "type": "output_text", "text": "..." }
      ]
    }
  ],
  "usage": { "input_tokens": 0, "output_tokens": 0, "total_tokens": 0 }
}

For reasoning models, output may contain a type: "reasoning" item before the message. Filter by type, or just use output_text.

StreamingCopy Anchor Link

Add "stream": true and the server returns SSE events. Incremental text is in the delta field of response.output_text.delta events, and the stream ends with a response.completed event:

event: response.output_text.delta
data: {"type":"response.output_text.delta","delta":"Hi", ...}

event: response.completed
data: {"type":"response.completed","response":{"status":"completed", ...}}

With an SDK, pass stream=True / stream: true and iterate over the events.

See the API Reference for all fields.