List of Supported Models

DeepIntShield currently registers 29 built-in provider identities. The live GET /v1/models result remains authoritative for the configured workspace: model availability depends on account access, region, provider changes, and per-key allow-lists.

Choose an operation for the model

Model names are preserved as opaque identifiers, including deployment aliases, repository paths, and version suffixes. Use provider/model-id for explicit routing. Listing a model does not establish support for every endpoint.

The dashboard’s model parameter endpoint, GET /api/models/parameters?provider=runway&model=gen4_image, supplies the metadata used by the Playground:

Field	Meaning
`provider_supported_operations`	Operations implemented by the gateway’s provider adapter.
`supported_operations`	Known model operations, restricted to the adapter’s implemented operations. Omitted means unknown; `[]` means no advertised supported operation.
`required_inputs`	Required inputs for each operation, such as a reference image or voice.
`operation_parameters`	Operation-specific controls and their supported values.
`unsupported_inputs`	Inputs excluded by the adapter, such as a mask for Runway image editing.
`parameters_unavailable`	Sampling metadata is unavailable; the model can still be selected.

This is a management endpoint and uses dashboard/API management authorization. Inference clients do not need to fetch it before making requests. Missing metadata does not add temperature, reasoning effort, or token-limit defaults; choose the operation explicitly for an unknown or custom model.

Runway’s inventory uses a bundled model snapshot plus explicitly configured IDs because the provider has no model-list endpoint. It does not verify remote account access. ElevenLabs discovery preserves speech-model capabilities; model parameter metadata distinguishes speech and supported transcription operations so audio models are offered for the appropriate operation.

Guides for all built-in provider identities

Wafer’s live model cards include model-specific capabilities and pricing. Other provider prices may come from the catalog or workspace overrides; see Model catalog and pricing. Resource operations that do not map cleanly to a model list are documented under Protocol operations.