openapi: 3.1.0
info:
  title: Cerebras Inference Chat Public Models API
  version: 1.0.0
  description: 'Generate conversational responses using a structured message format with roles (system, user, assistant, developer, tool). Best for chatbots, assistants, and multi-turn conversations.

    '
servers:
- url: https://api.cerebras.ai
  description: Cerebras Inference API
security:
- BearerAuth: []
tags:
- name: Public Models
paths:
  /public/v1/models:
    get:
      description: List all publicly available models (modelVisibility='PUBLIC' only). Use the `format` parameter to get responses compatible with OpenRouter or HuggingFace APIs.
      operationId: list_public_models_endpoint_public_v1_models_get
      parameters:
      - description: 'Output format: ''default'' (Cerebras schema), ''openrouter'', or ''huggingface'''
        in: query
        name: format
        required: false
        schema:
          $ref: '#/components/schemas/ModelFormat'
          default: default
          description: 'Output format: ''default'' (Cerebras schema), ''openrouter'', or ''huggingface'''
      - in: header
        name: X-Amz-Cf-Id
        required: false
        schema:
          anyOf:
          - type: string
          - type: 'null'
          title: X-Amz-Cf-Id
      - in: header
        name: CF-RAY
        required: false
        schema:
          anyOf:
          - type: string
          - type: 'null'
          title: Cf-Ray
      responses:
        '200':
          content:
            application/json:
              schema:
                anyOf:
                - $ref: '#/components/schemas/PublicModelsListResponse'
                - $ref: '#/components/schemas/OpenRouterModelsResponse'
                - $ref: '#/components/schemas/HuggingFaceModelsResponse'
                title: Response List Public Models Endpoint Public V1 Models Get
          description: Successful Response
        '422':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/HTTPValidationError'
          description: Validation Error
      summary: List all available models
      tags:
      - Public Models
  /public/v1/models/{model_id}:
    get:
      description: Retrieve a specific publicly available model by ID (modelVisibility='PUBLIC' only). Use the `format` parameter to get responses compatible with OpenRouter or HuggingFace APIs.
      operationId: get_public_model_public_v1_models__model_id__get
      parameters:
      - in: path
        name: model_id
        required: true
        schema:
          title: Model Id
          type: string
      - description: 'Output format: ''default'' (Cerebras schema), ''openrouter'', or ''huggingface'''
        in: query
        name: format
        required: false
        schema:
          $ref: '#/components/schemas/ModelFormat'
          default: default
          description: 'Output format: ''default'' (Cerebras schema), ''openrouter'', or ''huggingface'''
      - in: header
        name: X-Amz-Cf-Id
        required: false
        schema:
          anyOf:
          - type: string
          - type: 'null'
          title: X-Amz-Cf-Id
      - in: header
        name: CF-RAY
        required: false
        schema:
          anyOf:
          - type: string
          - type: 'null'
          title: Cf-Ray
      responses:
        '200':
          content:
            application/json:
              schema:
                anyOf:
                - $ref: '#/components/schemas/PublicModel'
                - $ref: '#/components/schemas/OpenRouterModel'
                - $ref: '#/components/schemas/HuggingFaceModel'
                title: Response Get Public Model Public V1 Models  Model Id  Get
          description: Successful Response
        '422':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/HTTPValidationError'
          description: Validation Error
      summary: Retrieve a specific model
      tags:
      - Public Models
components:
  schemas:
    OpenRouterPricing:
      description: OpenRouter pricing format.
      properties:
        completion:
          description: Cost per output token as string
          title: Completion
          type: string
        image:
          default: '0'
          description: Cost per image as string
          title: Image
          type: string
        input_cache_read:
          default: '0'
          description: Cost per cached input token read as string
          title: Input Cache Read
          type: string
        input_cache_write:
          default: '0'
          description: Cost per cached input token write as string
          title: Input Cache Write
          type: string
        prompt:
          description: Cost per input token as string
          title: Prompt
          type: string
        request:
          default: '0'
          description: Cost per request as string
          title: Request
          type: string
      required:
      - prompt
      - completion
      title: OpenRouterPricing
      type: object
    HuggingFacePricing:
      description: HuggingFace pricing format - price in USD per million tokens.
      properties:
        input:
          description: Price in USD per million input tokens
          title: Input
          type: number
        output:
          description: Price in USD per million output tokens
          title: Output
          type: number
      required:
      - input
      - output
      title: HuggingFacePricing
      type: object
    PublicModel:
      description: 'Complete model specification following OpenAI-compatible schema

        with extensions for OpenRouter/HuggingFace compatibility.'
      properties:
        architecture:
          $ref: '#/components/schemas/ModelArchitecture'
          description: Technical architecture details of the model.
        capabilities:
          $ref: '#/components/schemas/ModelCapabilities'
          description: The capabilities supported by the model.
        created:
          description: The Unix timestamp (in seconds) when the model was created.
          minimum: 0.0
          title: Created
          type: integer
        datacenter_locations:
          description: List of datacenter locations where this model is deployed (e.g., ['us-east-1', 'eu-west-1']).
          items:
            type: string
          title: Datacenter Locations
          type: array
        deprecated:
          default: false
          description: Indicates if the model is deprecated and should not be used for new applications.
          title: Deprecated
          type: boolean
        description:
          description: A brief description of the model.
          maxLength: 1000
          minLength: 1
          title: Description
          type: string
        hugging_face_id:
          anyOf:
          - type: string
          - type: 'null'
          description: The corresponding HuggingFace Hub model ID, if available (e.g., 'meta-llama/Llama-3.1-8B-Instruct').
          title: Hugging Face Id
        id:
          description: The unique identifier for the model (e.g., 'llama3.1-8b').
          maxLength: 200
          minLength: 1
          title: Id
          type: string
        limits:
          $ref: '#/components/schemas/ModelLimits'
          description: Usage limits and constraints for the model.
        name:
          description: The human-readable name of the model.
          minLength: 1
          title: Name
          type: string
        object:
          const: model
          default: model
          description: The object type, which is always 'model'.
          title: Object
          type: string
        owned_by:
          description: The organization that owns or created the model.
          minLength: 1
          title: Owned By
          type: string
        preview:
          default: false
          description: Indicates if the model is in preview or beta status.
          title: Preview
          type: boolean
        pricing:
          $ref: '#/components/schemas/ModelPricing'
          description: Pricing details for the model.
        quantization:
          anyOf:
          - type: string
          - type: 'null'
          description: Quantization precision (e.g., 'FP16', 'FP16/FP8 (weights only)').
          title: Quantization
        supported_parameters:
          $ref: '#/components/schemas/SupportedParameters'
      required:
      - id
      - created
      - owned_by
      - name
      - description
      - pricing
      - capabilities
      - supported_parameters
      - architecture
      - limits
      title: PublicModel
      type: object
    ModelCapabilities:
      description: Capabilities and features supported by the model.
      properties:
        function_calling:
          default: false
          description: Indicates if the model supports function calling (tool use).
          title: Function Calling
          type: boolean
        json_mode:
          default: false
          description: Indicates if the model supports JSON mode (guaranteed JSON output).
          title: Json Mode
          type: boolean
        parallel_tool_calls:
          default: false
          description: Indicates if the model supports parallel tool calls.
          title: Parallel Tool Calls
          type: boolean
        reasoning:
          default: false
          description: Indicates if the model supports reasoning/chain-of-thought outputs.
          title: Reasoning
          type: boolean
        response_format:
          default: false
          description: Indicates if the model supports the response_format parameter.
          title: Response Format
          type: boolean
        streaming:
          default: true
          description: Indicates if the model supports streaming responses via Server-Sent Events (SSE).
          title: Streaming
          type: boolean
        structured_outputs:
          default: false
          description: Indicates if the model supports structured outputs (e.g. JSON schema enforcement).
          title: Structured Outputs
          type: boolean
        tool_choice:
          default: false
          description: Indicates if the model supports the tool_choice parameter.
          title: Tool Choice
          type: boolean
        tools:
          default: false
          description: Indicates if the model supports the tools parameter.
          title: Tools
          type: boolean
        vision:
          default: false
          description: Indicates if the model accepts image inputs (vision capabilities).
          title: Vision
          type: boolean
      title: ModelCapabilities
      type: object
    HuggingFaceCapabilities:
      description: HuggingFace capabilities format.
      properties:
        function_calling:
          default: false
          title: Function Calling
          type: boolean
        streaming:
          default: true
          title: Streaming
          type: boolean
        structured_outputs:
          default: false
          title: Structured Outputs
          type: boolean
        vision:
          default: false
          title: Vision
          type: boolean
      title: HuggingFaceCapabilities
      type: object
    OpenRouterInfo:
      description: OpenRouter metadata.
      properties:
        slug:
          description: OpenRouter slug for the model
          title: Slug
          type: string
      required:
      - slug
      title: OpenRouterInfo
      type: object
    DatacenterLocation:
      description: Datacenter location information.
      properties:
        country_code:
          description: ISO 3166 Alpha-2 country code
          title: Country Code
          type: string
      required:
      - country_code
      title: DatacenterLocation
      type: object
    ModelLimits:
      description: Rate limits and constraints for the model.
      properties:
        max_completion_tokens:
          description: The maximum number of tokens that can be generated in a single completion.
          exclusiveMinimum: 0.0
          title: Max Completion Tokens
          type: integer
        max_context_length:
          description: The maximum context window size in tokens.
          exclusiveMinimum: 0.0
          title: Max Context Length
          type: integer
        requests_per_minute:
          anyOf:
          - minimum: 0.0
            type: integer
          - type: 'null'
          description: The default rate limit for requests per minute (RPM).
          title: Requests Per Minute
        tokens_per_minute:
          anyOf:
          - minimum: 0.0
            type: integer
          - type: 'null'
          description: The default rate limit for tokens per minute (TPM).
          title: Tokens Per Minute
      required:
      - max_context_length
      - max_completion_tokens
      title: ModelLimits
      type: object
    ModelArchitecture:
      description: Architecture details of the model.
      properties:
        instruct_type:
          anyOf:
          - type: string
          - type: 'null'
          description: The instruction format type used for fine-tuning (e.g., 'llama3', 'chatml').
          title: Instruct Type
        modality:
          description: The modality of the model (e.g., 'text', 'text+vision', 'multimodal').
          enum:
          - text
          - text+vision
          - multimodal
          title: Modality
          type: string
        tokenizer:
          description: The tokenizer used by the model (e.g., 'Llama3', 'GPT4').
          title: Tokenizer
          type: string
      required:
      - modality
      - tokenizer
      title: ModelArchitecture
      type: object
    PublicModelsListResponse:
      description: OpenAI-compatible list of models.
      properties:
        data:
          description: Array of model objects
          items:
            $ref: '#/components/schemas/PublicModel'
          title: Data
          type: array
        object:
          const: list
          default: list
          description: Object type (always 'list')
          title: Object
          type: string
      required:
      - data
      title: PublicModelsListResponse
      type: object
    OpenRouterModel:
      description: Model in OpenRouter-compatible format.
      properties:
        context_length:
          title: Context Length
          type: integer
        created:
          description: Unix timestamp when model was created
          title: Created
          type: integer
        datacenters:
          description: Datacenter locations
          items:
            $ref: '#/components/schemas/DatacenterLocation'
          title: Datacenters
          type: array
        description:
          default: ''
          description: Model description
          title: Description
          type: string
        hugging_face_id:
          default: ''
          description: The corresponding HuggingFace Hub model ID, if available
          title: Hugging Face Id
          type: string
        id:
          description: Model ID with provider prefix, e.g., 'cerebras/llama3.1-8b'
          title: Id
          type: string
        input_modalities:
          description: Supported input modalities (text, image, file)
          items:
            type: string
          title: Input Modalities
          type: array
        max_output_length:
          description: Maximum number of output tokens
          title: Max Output Length
          type: integer
        name:
          title: Name
          type: string
        openrouter:
          anyOf:
          - $ref: '#/components/schemas/OpenRouterInfo'
          - type: 'null'
          description: OpenRouter metadata
        output_modalities:
          description: Supported output modalities (text, image, file)
          items:
            type: string
          title: Output Modalities
          type: array
        pricing:
          $ref: '#/components/schemas/OpenRouterPricing'
        quantization:
          default: fp16
          description: Model quantization (fp16 only for Cerebras)
          title: Quantization
          type: string
        supported_features:
          description: List of supported features
          items:
            type: string
          title: Supported Features
          type: array
        supported_sampling_parameters:
          description: List of supported sampling parameters
          items:
            type: string
          title: Supported Sampling Parameters
          type: array
      required:
      - id
      - name
      - created
      - context_length
      - max_output_length
      - pricing
      title: OpenRouterModel
      type: object
    HuggingFaceModel:
      description: 'Model in HuggingFace-compatible format for inference providers.


        This format is used by HuggingFace to power their provider comparison table

        and provider selection features.'
      properties:
        capabilities:
          $ref: '#/components/schemas/HuggingFaceCapabilities'
        context_length:
          description: Supported context length in tokens
          title: Context Length
          type: integer
        created:
          title: Created
          type: integer
        hugging_face_id:
          anyOf:
          - type: string
          - type: 'null'
          description: The corresponding HuggingFace Hub model ID
          title: Hugging Face Id
        id:
          title: Id
          type: string
        object:
          const: model
          default: model
          title: Object
          type: string
        owned_by:
          title: Owned By
          type: string
        pricing:
          $ref: '#/components/schemas/HuggingFacePricing'
      required:
      - id
      - created
      - owned_by
      - context_length
      - pricing
      title: HuggingFaceModel
      type: object
    OpenRouterModelsResponse:
      description: OpenRouter-compatible list of models.
      properties:
        data:
          items:
            $ref: '#/components/schemas/OpenRouterModel'
          title: Data
          type: array
      required:
      - data
      title: OpenRouterModelsResponse
      type: object
    ValidationError:
      properties:
        loc:
          items:
            anyOf:
            - type: string
            - type: integer
          title: Location
          type: array
        msg:
          title: Message
          type: string
        type:
          title: Error Type
          type: string
      required:
      - loc
      - msg
      - type
      title: ValidationError
      type: object
    SupportedParameters:
      description: Sampling parameters supported by the model.
      properties:
        frequency_penalty:
          default: false
          description: Supports frequency_penalty parameter.
          title: Frequency Penalty
          type: boolean
        logit_bias:
          default: false
          description: Supports logit_bias parameter.
          title: Logit Bias
          type: boolean
        logprobs:
          default: false
          description: Supports logprobs output.
          title: Logprobs
          type: boolean
        max_completion_tokens:
          default: true
          description: Supports max_completion_tokens parameter.
          title: Max Completion Tokens
          type: boolean
        presence_penalty:
          default: false
          description: Supports presence_penalty parameter.
          title: Presence Penalty
          type: boolean
        repetition_penalty:
          default: false
          description: Supports repetition_penalty parameter.
          title: Repetition Penalty
          type: boolean
        seed:
          default: true
          description: Supports seed for reproducible outputs.
          title: Seed
          type: boolean
        stop:
          default: true
          description: Supports stop sequences parameter.
          title: Stop
          type: boolean
        temperature:
          default: true
          description: Supports temperature sampling parameter.
          title: Temperature
          type: boolean
        top_logprobs:
          default: false
          description: Supports top_logprobs parameter.
          title: Top Logprobs
          type: boolean
        top_p:
          default: true
          description: Supports top_p (nucleus) sampling parameter.
          title: Top P
          type: boolean
      title: SupportedParameters
      type: object
    HTTPValidationError:
      properties:
        detail:
          items:
            $ref: '#/components/schemas/ValidationError'
          title: Detail
          type: array
      title: HTTPValidationError
      type: object
    ModelPricing:
      description: Pricing information for a model.
      example:
        completion: '0.0000001'
        prompt: '0.0000001'
      properties:
        completion:
          description: Cost per token for completion (output) tokens in USD.
          pattern: ^(?!^[-+.]*$)[+-]?0*\d*\.?\d*$
          title: Completion
          type: string
        prompt:
          description: Cost per token for prompt (input) tokens in USD.
          pattern: ^(?!^[-+.]*$)[+-]?0*\d*\.?\d*$
          title: Prompt
          type: string
      required:
      - prompt
      - completion
      title: ModelPricing
      type: object
    HuggingFaceModelsResponse:
      description: HuggingFace-compatible list of models.
      properties:
        data:
          items:
            $ref: '#/components/schemas/HuggingFaceModel'
          title: Data
          type: array
        object:
          const: list
          default: list
          title: Object
          type: string
      required:
      - data
      title: HuggingFaceModelsResponse
      type: object
    ModelFormat:
      description: Output format for public models endpoint.
      enum:
      - default
      - openrouter
      - huggingface
      title: ModelFormat
      type: string
  securitySchemes:
    BearerAuth:
      type: http
      scheme: bearer
      description: 'API key for authentication. Obtain your key from the Cerebras Cloud console and pass it as `Authorization: Bearer YOUR_API_KEY`.

        '