Gengxi API Developer Docs

Responses API

Generate text, use tools, send multimodal input, and stream output with POST /v1/responses.

Create a response

bashCopy

curl https://api.gengxi.ai/v1/responses \
  -H "Authorization: Bearer $GENGXI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL",
    "input": "Hello"
  }'

Field	Purpose
model	A model name allowed for the current key and group
input	Text or structured multimodal input
instructions	High-level behavior for the response
stream	Return server-sent events when true

Generation controls

Common compatibility paths support or convert fields such as max_output_tokens, temperature, top_p, tools, tool_choice, and reasoning.effort.

jsonCopy

{
  "model": "YOUR_MODEL",
  "instructions": "Answer concisely.",
  "input": "Compare queues and streams.",
  "reasoning": {"effort": "medium"},
  "max_output_tokens": 800
}

Tools and multimodal input

Requests can include compatible tool definitions and multimodal input. Exact input types and tool behavior depend on the selected model and route.

Treat reasoning.summary, include, store, parallel_tool_calls, text.format, text.verbosity, service_tier, prompt_cache_key, previous_response_id, and metadata as route-dependent. Gengxi API does not claim support for every field in every OpenAI Responses implementation.