https://platform.r3al.ai. All endpoints below begin with /v1.
First create a project and complete model setup. An API key authenticates your requests; it does not provision a model or start compute.
Authentication
Use a team account’s API key for inference.
Chat completions
Request a response or a stream from your model.
Runtime and demos
Inspect readiness and manage a configured session.
Projects
Reference for the session-authenticated workspace.
Authentication boundaries
The inference endpoint uses a chat-completions-shaped request and response. It is not a complete OpenAI API implementation: tools, file inputs, and arbitrary model routing are not supported. Streaming responses use server-sent events; incremental delivery depends on the connected runtime.
Existing vision quantization APIs remain in the legacy SDK documentation. They are separate from this project-based inference workflow.