API

The meinGPT API brings model responses and platform capabilities into your own applications. An OpenAI-compatible gateway provides endpoints for chat completions, embeddings, models, and transcription. Existing compatible clients can work with one centrally issued API key.

The calling application sends context and model choice and processes the result in its own interface. Authentication uses a protected bearer token. Assistants and workflows can also be called programmatically when their platform logic should be reused. The meinGPT gateway runs in the EU; further processing depends on the chosen model and deployment option.

OpenAI-compatible does not mean every provider parameter is supported identically. Error handling, timeouts, validation, and business approval remain the responsibility of the integrating application. If an assistant instead needs access to functions in your system, describe them as bounded tools through MCP.