Felo API PlatformFelo API Platform
v0.3.0
OpenAPI 3.1.0

Schemas

Felo API Support
MIT
Server:https://openapi.felo.ai
Client Libraries
Shell Curl

Schemas

  • created
    Type: integer Format: int64
    required

    Model catalog creation timestamp in Unix seconds.

  • id
    Type: string
    required

    Felo model id.

  • object
    const:  
    model
    required
  • owned_by
    Type: string
    required
  • model_kind
    Type: string

    Effective model capability category after enabled provider mapping overrides: chat, image_generation, video_generation, or decisions. Use the calling endpoint matching this category.

  • propertyName
    Type: anything
  • data
    Type: array object[] ·
    required
  • object
    const:  
    list
    required
  • propertyName
    Type: anything

OpenAI-compatible error envelope. The Router may include code and message without all optional upstream fields.

  • error
    Type: object
    required
  • propertyName
    Type: anything
  • messages
    Type: array object[] · 1…
    required
  • model
    Type: string
    required
  • max_completion_tokens
    Type: integer
    min:  
    1

    Integer numbers.

  • max_tokens
    Type: integer
    min:  
    1

    Integer numbers.

  • n
    Type: integer
    min:  
    1

    Integer numbers.

  • response_format
    Type: object
  • stop
    nullable
    • Type: string
  • stream
    Type: boolean
  • temperature
    Type: number
    min:  
    0
  • tool_choice
    • Type: string
  • tools
    Type: array object[]
  • top_p
    Type: number
    min:  
    0
    max:  
    1
  • user
    Type: string
  • propertyName
    Type: anything
  • content
    nullable
    required
    • Type: string
  • role
    Type: string enum
    required
    values
    • system
    • developer
    • user
    • assistant
    • tool
  • name
    Type: string
  • tool_call_id
    Type: string
  • tool_calls
    Type: array object[]
  • propertyName
    Type: anything
  • choices
    Type: array object[] ·
    required
  • id
    Type: string
    required
  • object
    Type: string
    required

    Usually chat.completion for non-streaming responses.

  • created
    Type: integer Format: int64

    Signed 64-bit integers (long type).

  • model
    Type: string
  • system_fingerprint
    Type: string
  • usage
    Type: object ·

    Provider usage object. Fields vary by endpoint and provider.

  • propertyName
    Type: anything
  • finish_reason
    Type: string | null
  • index
    Type: integer

    Integer numbers.

  • logprobs
    Type: object | null
  • message
    Type: object ·
  • propertyName
    Type: anything

Provider usage object. Fields vary by endpoint and provider.

  • propertyName
    Type: anything
  • input
    required
    • Type: string
  • model
    Type: string
    required
  • include
    Type: array string[]
  • instructions
    Type: string
  • max_output_tokens
    Type: integer
    min:  
    1

    Integer numbers.

  • metadata
    Type: object
  • previous_response_id
    Type: string
  • reasoning
    Type: object
  • store
    Type: boolean
  • stream
    Type: boolean
  • temperature
    Type: number
    min:  
    0
  • text
    Type: object
  • tool_choice
    • Type: string
  • tools
    Type: array object[]
  • top_p
    Type: number
    min:  
    0
    max:  
    1
  • propertyName
    Type: anything
  • id
    Type: string
    required
  • object
    const:  
    response
    required
  • status
    Type: string enum
    required
    values
    • completed
    • failed
    • in_progress
    • cancelled
    • queued
    • incomplete
  • error
    Type: object ·
  • metadata
    Type: object
  • model
    Type: string
  • output
    Type: array object[]
  • output_text
    Type: string
  • usage
    Type: object ·

    Provider usage object. Fields vary by endpoint and provider.

  • propertyName
    Type: anything
  • code
    Type: string
    required
  • message
    Type: string
    required
  • propertyName
    Type: anything
  • max_tokens
    Type: integer
    min:  
    1
    required

    Integer numbers.

  • messages
    Type: array object[] · 1…
    required
  • model
    Type: string
    required
  • metadata
    Type: object
  • service_tier
    Type: string
  • stop_sequences
    Type: array string[]
  • stream
    Type: boolean
  • system
    • Type: string
  • temperature
    Type: number
    min:  
    0
    max:  
    1
  • thinking
    Type: object
  • tool_choice
    • Type: string
  • tools
    Type: array object[]
  • top_k
    Type: integer
    min:  
    0

    Integer numbers.

  • top_p
    Type: number
    min:  
    0
    max:  
    1
  • propertyName
    Type: anything
  • content
    required
    • Type: string
  • role
    Type: string enum
    required
    values
    • user
    • assistant
  • propertyName
    Type: anything
  • type
    Type: string enum
    required
    values
    • text
    • image
    • document
    • tool_use
    • tool_result
    • thinking
  • content
    Type: string | array
  • id
    Type: string
  • input
    Type: object
  • name
    Type: string
  • source
    Type: object
  • text
    Type: string
  • propertyName
    Type: anything
  • content
    Type: array object[] ·
    required
  • id
    Type: string
    required
  • model
    Type: string
    required
  • role
    const:  
    assistant
    required
  • stop_reason
    Type: string | null
    required
  • type
    const:  
    message
    required
  • stop_sequence
    Type: string | null
  • usage
    Type: object ·

    Provider usage object. Fields vary by endpoint and provider.

  • propertyName
    Type: anything

Claude Errors API-compatible error envelope for Messages requests.

  • error
    Type: object
    required
  • type
    const:  
    error
    required
  • request_id
    Type: string
  • propertyName
    Type: anything

Video generation input. Either flattened ({model, prompt, ...}) or nested ({model, input: {...}}). input takes precedence. Exactly one input mode must be used; mixing first/last-frame with multimodal reference inputs is rejected.

  • model
    Type: string
    required

    Felo-facing model name, e.g. seedance-2.0-mini.

  • aspect_ratio
    Type: string enum
    values
    • 16:9
    • 9:16
    • 1:1
    • 4:3
    • 3:4
    • 21:9
    • adaptive
  • duration
    Type: integer
    min:  
    4
    max:  
    15

    Integer numbers.

  • first_frame_url
    Type: string Format: uri
  • generate_audio
    Type: boolean
  • input
    Type: object

    Optional nested input object; when present, its fields take precedence over sibling fields.

  • last_frame_url
    Type: string Format: uri
  • prompt
    Type: string
    max length:  
    2000
  • reference_audio_urls
    Type: array string[] …3Format: uri
  • reference_image_urls
    Type: array string[] …9Format: uri
  • reference_video_urls
    Type: array string[] …3Format: uri
  • resolution
    Type: string enum
    values
    • 480p
    • 720p
    • 1080p
    • 4k
  • kind
    enum
    const:  
    video
    required
    values
    • video
  • outputs
    Type: array string[]
    required

    Empty until the task succeeds.

  • status
    enum
    const:  
    queued
    required
    values
    • queued
  • task_id
    Type: string
    required

    Task id to use for polling.

  • kind
    enum
    const:  
    video
    required
    values
    • video
  • outputs
    Type: array string[] Format: uri
    required
  • status
    Type: string enum
    required
    values
    • queued
    • running
    • succeeded
    • failed
    • interrupted
  • task_id
    Type: string
    required
  • error
    Type: object

    Present when the task failed.

OpenAI-compatible error envelope.

  • error
    Type: object
Schemas — Model API