ModelDescriptor
HTTP API schema
One model offered for one capability interface, with the configuration it supports and how to prompt it. Exactly the section that matches the capability interface is present (`speech`, `speechToSpeech`, `image` or `music`).
ModelDescriptor fields
- capabilityInterface
stringRequired Canonical name of the capability interface the model serves, for example port1/systems/zero/capability-interfaces/speech-audio-generation.
- model
stringRequired The model identifier to pass as `model` in the interface generation message.
- displayName
stringRequired Human readable model name.
- promptingInstructions
stringRequired Free text on how to prompt the model, for example whether lyrics go in the prompt or in a separate field and which emotion or sound tags the model understands.
- maxPromptCharacters
integer Longest accepted text or prompt in characters, when the vendor documents one.
minimum: 1
- provenanceModelProvenanceRequired
provenance: Reference ModelProvenance
- speechModelSpeechConfiguration
speech: Reference ModelSpeechConfiguration
- speechToSpeechModelSpeechToSpeechConfiguration
speechToSpeech: Reference ModelSpeechToSpeechConfiguration
- imageModelImageConfiguration
image: Reference ModelImageConfiguration
- musicModelMusicConfiguration
music: Reference ModelMusicConfiguration