Create an OpenAI inference endpoint Added in 8.13.0

PUT /_inference/{task_type}/{elasticsearch_inference_id}
Api key auth

Create an inference endpoint to perform an inference task with the elasticsearch service.


Your Elasticsearch deployment contains preconfigured ELSER and E5 inference endpoints, you only need to create the enpoints using the API if you want to customize the settings.

If you use the ELSER or the E5 model through the elasticsearch service, the API request will automatically download and deploy the model if it isn't downloaded yet.


You might see a 502 bad gateway error in the response when using the Kibana Console. This error usually just reflects a timeout, while the model downloads in the background. You can check the download progress in the Machine Learning UI. If using the Python client, you can set the timeout parameter to a higher value.

After creating the endpoint, wait for the model deployment to complete before using it. To verify the deployment status, use the get trained model statistics API. Look for "state": "fully_allocated" in the response and ensure that the "allocation_count" matches the "target_allocation_count". Avoid creating multiple endpoints for the same model unless required, as each endpoint consumes significant resources.

Path parameters

  • task_type string Required

    The type of the inference task that the model will perform.

    Values are rerank, sparse_embedding, or text_embedding.

  • elasticsearch_inference_id string Required

    The unique identifier of the inference endpoint. The must not match the model_id.

application/json

Body

  • chunking_settings object
    Hide chunking_settings attributes Show chunking_settings attributes object
    • service string Required

      The service type

    • service_settings object Required
    • task_settings object
    • max_chunk_size number

      The maximum size of a chunk in words. This value cannot be higher than 300 or lower than 20 (for sentence strategy) or 10 (for word strategy).

    • overlap number

      The number of overlapping words for chunks. It is applicable only to a word chunking strategy. This value cannot be higher than half the max_chunk_size value.

    • sentence_overlap number

      The number of overlapping sentences for chunks. It is applicable only for a sentence chunking strategy. It can be either 1 or 0.

    • strategy string

      The chunking strategy: sentence or word.

  • service string Required

    Value is elasticsearch.

  • service_settings object Required
    Hide service_settings attributes Show service_settings attributes object
    • adaptive_allocations object
      Hide adaptive_allocations attributes Show adaptive_allocations attributes object
      • enabled boolean

        Turn on adaptive_allocations.

      • max_number_of_allocations number

        The maximum number of allocations to scale to. If set, it must be greater than or equal to min_number_of_allocations.

      • min_number_of_allocations number

        The minimum number of allocations to scale to. If set, it must be greater than or equal to 0. If not defined, the deployment scales to 0.

    • deployment_id string

      The deployment identifier for a trained model deployment. When deployment_id is used the model_id is optional.

    • model_id string Required

      The name of the model to use for the inference task. It can be the ID of a built-in model (for example, .multilingual-e5-small for E5) or a text embedding model that was uploaded by using the Eland client.

      External documentation
    • num_allocations number

      The total number of allocations that are assigned to the model across machine learning nodes. Increasing this value generally increases the throughput. If adaptive allocations are enabled, do not set this value because it's automatically set.

    • num_threads number Required

      The number of threads used by each model allocation during inference. This setting generally increases the speed per inference request. The inference process is a compute-bound process; threads_per_allocations must not exceed the number of available allocated processors per node. The value must be a power of 2. The maximum value is 32.

  • task_settings object
    Hide task_settings attribute Show task_settings attribute object
    • return_documents boolean

      For a rerank task, return the document instead of only the index.

Responses

  • 200 application/json
    Hide response attributes Show response attributes object
    • Hide attributes Show attributes object
      • max_chunk_size number

        The maximum size of a chunk in words. This value cannot be higher than 300 or lower than 20 (for sentence strategy) or 10 (for word strategy).

      • overlap number

        The number of overlapping words for chunks. It is applicable only to a word chunking strategy. This value cannot be higher than half the max_chunk_size value.

      • sentence_overlap number

        The number of overlapping sentences for chunks. It is applicable only for a sentence chunking strategy. It can be either 1 or 0.

      • strategy string

        The chunking strategy: sentence or word.

    • service string Required

      The service type

    • service_settings object Required
    • task_settings object
    • inference_id string Required

      The inference Id

    • task_type string Required

      Values are sparse_embedding, text_embedding, rerank, completion, or chat_completion.
PUT /_inference/{task_type}/{elasticsearch_inference_id} 
curl \
 --request PUT 'http://api.example.com/_inference/{task_type}/{elasticsearch_inference_id}' \
 --header "Authorization: $API_KEY" \
 --header "Content-Type: application/json" \
 --data '"{\n    \"service\": \"elasticsearch\",\n    \"service_settings\": {\n        \"adaptive_allocations\": { \n        \"enabled\": true,\n        \"min_number_of_allocations\": 1,\n        \"max_number_of_allocations\": 4\n        },\n        \"num_threads\": 1,\n        \"model_id\": \".elser_model_2\" \n    }\n}"'
Request examples
Run `PUT _inference/sparse_embedding/my-elser-model` to create an inference endpoint that performs a `sparse_embedding` task. The `model_id` must be the ID of one of the built-in ELSER models. The API will automatically download the ELSER model if it isn't already downloaded and then deploy the model. 
{
    "service": "elasticsearch",
    "service_settings": {
        "adaptive_allocations": { 
        "enabled": true,
        "min_number_of_allocations": 1,
        "max_number_of_allocations": 4
        },
        "num_threads": 1,
        "model_id": ".elser_model_2" 
    }
}
Run `PUT _inference/rerank/my-elastic-rerank` to create an inference endpoint that performs a rerank task using the built-in Elastic Rerank cross-encoder model. The `model_id` must be `.rerank-v1`, which is the ID of the built-in Elastic Rerank model. The API will automatically download the Elastic Rerank model if it isn't already downloaded and then deploy the model. Once deployed, the model can be used for semantic re-ranking with a `text_similarity_reranker` retriever. 
{
    "service": "elasticsearch",
    "service_settings": {
        "model_id": ".rerank-v1", 
        "num_threads": 1,
        "adaptive_allocations": { 
        "enabled": true,
        "min_number_of_allocations": 1,
        "max_number_of_allocations": 4
        }
    }
}
Run `PUT _inference/text_embedding/my-e5-model` to create an inference endpoint that performs a `text_embedding` task. The `model_id` must be the ID of one of the built-in E5 models. The API will automatically download the E5 model if it isn't already downloaded and then deploy the model. 
{
    "service": "elasticsearch",
    "service_settings": {
        "num_allocations": 1,
        "num_threads": 1,
        "model_id": ".multilingual-e5-small" 
    }
}
Run `PUT _inference/text_embedding/my-msmarco-minilm-model` to create an inference endpoint that performs a `text_embedding` task with a model that was uploaded by Eland. 
{
    "service": "elasticsearch",
    "service_settings": {
        "num_allocations": 1,
        "num_threads": 1,
        "model_id": "msmarco-MiniLM-L12-cos-v5" 
    }
}
Run `PUT _inference/text_embedding/my-e5-model` to create an inference endpoint that performs a `text_embedding` task and to configure adaptive allocations. The API request will automatically download the E5 model if it isn't already downloaded and then deploy the model. 
{
    "service": "elasticsearch",
    "service_settings": {
        "adaptive_allocations": {
        "enabled": true,
        "min_number_of_allocations": 3,
        "max_number_of_allocations": 10
        },
        "num_threads": 1,
        "model_id": ".multilingual-e5-small"
    }
}
Run `PUT _inference/sparse_embedding/use_existing_deployment` to use an already existing model deployment when creating an inference endpoint. 
{
    "service": "elasticsearch",
    "service_settings": {
        "deployment_id": ".elser_model_2"
    }
}
Response examples (200)
A successful response from `PUT _inference/sparse_embedding/use_existing_deployment`. It contains the model ID and the threads and allocations settings from the model deployment. 
{
  "inference_id": "use_existing_deployment",
  "task_type": "sparse_embedding",
  "service": "elasticsearch",
  "service_settings": {
    "num_allocations": 2,
    "num_threads": 1,
    "model_id": ".elser_model_2",
    "deployment_id": ".elser_model_2"
  },
  "chunking_settings": {
    "strategy": "sentence",
    "max_chunk_size": 250,
    "sentence_overlap": 1
  }
}