Add a LoRA adapter to an endpoint

Adds a LoRA adapter model to a dedicated endpoint. After this call, inference requests to the adapter model name will be routed to the specified endpoint. The endpoint must have LoRA enabled, and the adapter's base model must be compatible with the endpoint's model. The endpoint name prefix in model_id must match the resolved endpoint.

post/endpoints/{endpointId}/adapters

Request body

model_idstring required

Combined identifier in format "endpoint_name:adapter_model_name".

Example request

{
  "model_id": "username/Meta-Llama-3.1-8B-Instruct-def456:username/my-adapter-abc123"
}

Response

Adapter successfully bound to endpoint

model_idstring

Changes

No recorded changes to this endpoint across all 14 revisions of this API.