Skip to main content

Inference API

The Resultity network provides an OpenAI-compatible API for executing inference requests on decentralized infrastructure.
Users can integrate the service into their existing applications with minimal changes — simply replace the base URL and provide an API key.

This API is available via cloud.resultity.com and uses standard HTTPS requests.
Each request consumes internal credits, and usage is tracked per API key and namespace.

OpenAI Compatibility​

Resultity aims for maximum compatibility with the OpenAI API, making it easy to switch without changing client-side code.

Supported endpoints:

  • /v1/chat/completions — for chat-based models;
  • /v1/completions — for traditional text completion;
  • /v1/embeddings — vector embedding of input texts;
  • /v1/images/generations — image generation (planned);
  • /v1/audio/transcriptions — audio-to-text (planned);
  • /v1/audio/speech — text-to-speech (planned);
  • /v1/moderations — content safety and classification (planned).

Supported Parameters​

Requests use standard OpenAI-style payloads, with parameters including:

  • model: model name;
  • messages or prompt: input content;
  • temperature, top_p: sampling controls;
  • max_tokens, stop: generation limits;
  • stream, logprobs, n: advanced options (partially supported).

Resultity Cloud may not support all OpenAI models.
The list of available models is dynamically retrieved via the /v1/models endpoint and may differ depending on the active node fleet and region.

Compatibility Notes​

  • Most SDKs and wrappers designed for OpenAI can be reused with Resultity;
  • Some advanced parameters (e.g. function_call, tool_choice) may have limited support at launch;
  • Model names may differ slightly (e.g. rtity-chat-7b, rtity-embed-base), but aliases are available for compatibility;
  • Future versions may introduce Resultity-specific extensions to the standard.

This compatibility layer allows developers to quickly adopt decentralized inference while retaining the ecosystem benefits of OpenAI's API format.