Hyperbolic Discover Model And Chat

arazzo: 1.0.1
info:
  title: Hyperbolic Discover Model And Chat
  summary: List the live model catalog, pick a chat model, and run a chat completion against it.
  description: >-
    A foundational Hyperbolic Serverless Inference flow. The workflow first
    pulls the live model catalog so a caller can confirm a chat model is
    available, then sends a chat completion request to that model. Every step
    inlines its request — including the inline Authorization Bearer credential —
    so the flow can be read and executed without opening the underlying OpenAPI
    descriptions.
  version: 1.0.0
sourceDescriptions:
- name: modelsApi
  url: ../openapi/hyperbolic-models-api-openapi.yml
  type: openapi
- name: chatCompletionsApi
  url: ../openapi/hyperbolic-chat-completions-api-openapi.yml
  type: openapi
workflows:
- workflowId: discover-model-and-chat
  summary: Resolve an available chat model from the catalog and generate a chat completion.
  description: >-
    Calls the OpenAI-compatible models endpoint to confirm the live catalog,
    then submits a chat completion using a model id the caller supplies. The
    catalog read lets the caller validate the model exists before spending
    inference credits.
  inputs:
    type: object
    required:
    - apiKey
    - model
    - prompt
    properties:
      apiKey:
        type: string
        description: Hyperbolic API key passed as a Bearer token.
      model:
        type: string
        description: >-
          Chat model identifier (e.g. deepseek-ai/DeepSeek-V3 or
          meta-llama/Meta-Llama-3.1-70B-Instruct).
      systemPrompt:
        type: string
        description: Optional system instruction for the assistant.
      prompt:
        type: string
        description: The user message to send to the chat model.
      maxTokens:
        type: integer
        description: Maximum number of tokens to generate.
      temperature:
        type: number
        description: Sampling temperature between 0 and 2.
  steps:
  - stepId: listCatalog
    description: >-
      List all models currently available for inference so the target chat
      model can be confirmed before a completion is requested.
    operationId: listModels
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      models: $response.body#/data
      firstModelId: $response.body#/data/0/id
  - stepId: chat
    description: >-
      Send a single-turn chat completion to the supplied model and capture the
      assistant message text and token usage.
    operationId: createChatCompletion
    parameters:
    - name: Authorization
      in: header
      value: "Bearer $inputs.apiKey"
    requestBody:
      contentType: application/json
      payload:
        model: $inputs.model
        messages:
        - role: system
          content: $inputs.systemPrompt
        - role: user
          content: $inputs.prompt
        max_tokens: $inputs.maxTokens
        temperature: $inputs.temperature
        stream: false
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      completionId: $response.body#/id
      answer: $response.body#/choices/0/message/content
      finishReason: $response.body#/choices/0/finish_reason
      totalTokens: $response.body#/usage/total_tokens
  outputs:
    models: $steps.listCatalog.outputs.models
    answer: $steps.chat.outputs.answer
    totalTokens: $steps.chat.outputs.totalTokens