API REFERENCE

API reference

API origin: https://api.profundoai.com. Authenticate with a Bearer API key.

Endpoints

Chat completions

POST /v1/chat/completions
Authorization: Bearer $PROFUNDO_API_KEY
Content-Type: application/json

{"model":"glm-5.2","messages":[{"role":"user","content":"Hello"}]}

Required fields are model and messages. Confirmed fields include stream, temperature, max_tokens, top_p, frequency_penalty, presence_penalty, n, stop, seed, user, reasoning_effort, response_format, tools, tool_choice, parallel_tool_calls, top_logprobs, stream_options, metadata, and service_tier. Non-streaming responses are JSON; streaming uses server-sent events.

Responses

POST /v1/responses
Authorization: Bearer $PROFUNDO_API_KEY
Content-Type: application/json

{"model":"glm-5.2","input":"Hello"}

Profundo supports both Chat Completions and the Responses API. Responses require model and string or ordered-message-array input. Optional fields include instructions, stream, temperature, top_p, max_output_tokens, tools, tool_choice, reasoning, and metadata. Responses accepts tool definitions and tool choice, but tool-call output compatibility is not yet supported. Use Chat Completions for tool-using applications. Most models are text-only; MiniMax M3 also accepts image input. No model accepts audio, embeddings, or other multimodal content beyond what is documented.

Errors