AIAssistantsLLMEmbeddingBundle
One bundle entry grouped around a single embedding model, including all validated reasoning-model combinations for that embedding model.
{
"default": false,
"embedding_model_id": "string",
"embedding_model": {
"model_id": "string",
"display_name": "string",
"model_name": "string",
"sdk_provider": "string",
"expiry_time": 0
},
"supported_models": [
{
"default": false,
"display_name": "string",
"high_iq": {
"model_id": "string",
"display_name": "string",
"model_name": "string",
"sdk_provider": "string",
"expiry_time": 0
},
"normal_iq": {
"model_id": "string",
"display_name": "string",
"model_name": "string",
"sdk_provider": "string",
"expiry_time": 0
},
"low_iq": {
"model_id": "string",
"display_name": "string",
"model_name": "string",
"sdk_provider": "string",
"expiry_time": 0
}
}
],
"latest_bundle_versions": {
"latest_bundle_versions": "string"
},
"regions": [
"string"
]
}True when this is the system-level default embedding bundle. Only one entry across the entire catalog may be true.
Unique identifier for the bundle following the {provider}_{model_name} convention (e.g. gemini-gemini-embedding-001). Not intended for display in the UI; use embedding_model.display_name instead.
A single model reference used for both embedding models and reasoning-model models. model_id and model_name are the machine-readable identifiers; display_name is the UI label only.
All reasoning-model sets validated against this embedding model. The array is unordered; there is no positional index. Use the default flag or match by model_id to select an entry.
Most recently available bundle version per AI Assistant feature (firewall, avi).
Deployment regions supported for the embedding bundle. For GCP-backed bundles the operator selects one of these regions. For other providers this list is empty because the operator supplies the endpoint URL directly via AIAssistantsCustomModelConfig. Show a region picker only when this list is non-empty.