AI RAG Configuration
Configuration reference for the AI RAG plugin (`ai-rag`): the plugin parameters and request body parameters, each attribute with its type, whether it is required, its default and its accepted values.
Plugin Parameters
See plugin common configurations (opens in Apache APISIX docs) for configuration options available to all plugins.
embeddings_provider — object · required · since APISIX 3.12.0 · API7 Gateway 3.7.0
Embedding model provider configurations.
azure_openai — object · required · since APISIX 3.12.0 · API7 Gateway 3.7.0
Azure OpenAI embedding model configurations.
endpoint — string · required
Azure OpenAI embedding model endpoint.
api_key — string · required
Azure OpenAI API key. The value is encrypted with AES before being stored in etcd.
vector_search_provider — object · required · since APISIX 3.12.0 · API7 Gateway 3.7.0
Vector search provider configurations.
azure_ai_search — object · required · since APISIX 3.12.0 · API7 Gateway 3.7.0
Configurations of Azure AI Search.
endpoint — string · required
Azure AI Search endpoint.
api_key — string · required
Azure AI Search API key. The value is encrypted with AES before being stored in etcd.
ssl_verify — boolean · optional · default: true · since APISIX 3.17.0 · API7 Gateway 3.9.10
If true, verify TLS certificates when calling the embedding and vector search endpoints. Available in API7 Enterprise from version 3.9.10 and APISIX from version 3.17.0.
Request Body Parameters
The request body should follow the below configurations.
ai_rag — object · required
Request body RAG specifications.
embeddings — object · required
Embedding specifications.
The following parameters are available if you are working with Azure OpenAI.
input — string · required
Input prompt to the LLM, which will be used to compute embeddings and generate a RAG-enhanced response.
user — string · optional
A unique identifier representing your end user, which helps in monitoring and detecting abuse.
encoding_format — string · optional · default: float
Valid values: float or base64
Data type of the returned embeddings.
dimensions — integer · optional
Dimensions limit of the embedding model used to output the vectors. It should match the dimension of your embedding model. For instance, the dimensions for text-embedding-ada-002 are fixed at 1536. For text-embedding-3-small or text-embedding-3-large, dimensions range from 1 to 1536 and 3072, respectively.
vector_search — object · required
Vector search specifications.
The following parameter is available if you are working with Azure AI Search.
fields — string · required
Fields for the vector search.