Implement OpenShift AI integration for chat completion, embeddings, and reranking (#136624)

Author: Jan-Kazlouski-elastic

docs/changelog/136624.yaml

@@ -0,0 +1,5 @@
+pr: 136624
+summary: Added OpenShift AI text_embedding, completion, chat_completion and rerank support to the Inference Plugin
+area: Machine Learning
+type: enhancement
+issues: []

server/src/main/resources/transport/definitions/referable/ml_inference_openshift_ai_added.csv

@@ -0,0 +1 @@
+9218000

server/src/main/resources/transport/upper_bounds/9.3.csv

@@ -1 +1 @@
-resharding_shard_summary_in_esql,9217000
+ml_inference_openshift_ai_added,9218000

x-pack/plugin/inference/qa/inference-service-tests/src/javaRestTest/java/org/elasticsearch/xpack/inference/InferenceGetServicesIT.java

@@ -70,6 +70,7 @@ public void testGetServicesWithoutTaskType() throws IOException {
                     "llama",
                     "mistral",
                     "openai",
+                    "openshift_ai",
                     "streaming_completion_test_service",
                     "completion_test_service",
                     "test_reranking_service",
@@ -116,6 +117,7 @@ public void testGetServicesWithTextEmbeddingTaskType() throws IOException {
                     "llama",
                     "mistral",
                     "openai",
+                    "openshift_ai",
                     "text_embedding_test_service",
                     "voyageai",
                     "watsonxai"
@@ -140,6 +142,7 @@ public void testGetServicesWithRerankTaskType() throws IOException {
                     "elasticsearch",
                     "googlevertexai",
                     "jinaai",
+                    "openshift_ai",
                     "test_reranking_service",
                     "voyageai",
                     "hugging_face",
@@ -167,6 +170,7 @@ public void testGetServicesWithCompletionTaskType() throws IOException {
                     "googleaistudio",
                     "googlevertexai",
                     "openai",
+                    "openshift_ai",
                     "streaming_completion_test_service",
                     "completion_test_service",
                     "hugging_face",
@@ -188,6 +192,7 @@ public void testGetServicesWithChatCompletionTaskType() throws IOException {
                     "deepseek",
                     "elastic",
                     "openai",
+                    "openshift_ai",
                     "streaming_completion_test_service",
                     "hugging_face",
                     "amazon_sagemaker",

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/InferenceNamedWriteablesProvider.java

@@ -115,6 +115,10 @@
 import org.elasticsearch.xpack.inference.services.openai.completion.OpenAiChatCompletionTaskSettings;
 import org.elasticsearch.xpack.inference.services.openai.embeddings.OpenAiEmbeddingsServiceSettings;
 import org.elasticsearch.xpack.inference.services.openai.embeddings.OpenAiEmbeddingsTaskSettings;
+import org.elasticsearch.xpack.inference.services.openshiftai.completion.OpenShiftAiChatCompletionServiceSettings;
+import org.elasticsearch.xpack.inference.services.openshiftai.embeddings.OpenShiftAiEmbeddingsServiceSettings;
+import org.elasticsearch.xpack.inference.services.openshiftai.rerank.OpenShiftAiRerankServiceSettings;
+import org.elasticsearch.xpack.inference.services.openshiftai.rerank.OpenShiftAiRerankTaskSettings;
 import org.elasticsearch.xpack.inference.services.sagemaker.model.SageMakerModel;
 import org.elasticsearch.xpack.inference.services.sagemaker.schema.SageMakerSchemas;
 import org.elasticsearch.xpack.inference.services.settings.DefaultSecretSettings;
@@ -172,6 +176,7 @@ public static List<NamedWriteableRegistry.Entry> getNamedWriteables() {
         addCustomNamedWriteables(namedWriteables);
         addLlamaNamedWriteables(namedWriteables);
         addAi21NamedWriteables(namedWriteables);
+        addOpenShiftAiNamedWriteables(namedWriteables);
 
         addUnifiedNamedWriteables(namedWriteables);
 
@@ -446,6 +451,33 @@ private static void addOpenAiNamedWriteables(List<NamedWriteableRegistry.Entry>
         );
     }
 
+    private static void addOpenShiftAiNamedWriteables(List<NamedWriteableRegistry.Entry> namedWriteables) {
+        namedWriteables.add(
+            new NamedWriteableRegistry.Entry(
+                ServiceSettings.class,
+                OpenShiftAiEmbeddingsServiceSettings.NAME,
+                OpenShiftAiEmbeddingsServiceSettings::new
+            )
+        );
+        namedWriteables.add(
+            new NamedWriteableRegistry.Entry(
+                ServiceSettings.class,
+                OpenShiftAiChatCompletionServiceSettings.NAME,
+                OpenShiftAiChatCompletionServiceSettings::new
+            )
+        );
+        namedWriteables.add(
+            new NamedWriteableRegistry.Entry(
+                ServiceSettings.class,
+                OpenShiftAiRerankServiceSettings.NAME,
+                OpenShiftAiRerankServiceSettings::new
+            )
+        );
+        namedWriteables.add(
+            new NamedWriteableRegistry.Entry(TaskSettings.class, OpenShiftAiRerankTaskSettings.NAME, OpenShiftAiRerankTaskSettings::new)
+        );
+    }
+
     private static void addHuggingFaceNamedWriteables(List<NamedWriteableRegistry.Entry> namedWriteables) {
         namedWriteables.add(
             new NamedWriteableRegistry.Entry(

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/InferencePlugin.java

@@ -164,6 +164,7 @@
 import org.elasticsearch.xpack.inference.services.llama.LlamaService;
 import org.elasticsearch.xpack.inference.services.mistral.MistralService;
 import org.elasticsearch.xpack.inference.services.openai.OpenAiService;
+import org.elasticsearch.xpack.inference.services.openshiftai.OpenShiftAiService;
 import org.elasticsearch.xpack.inference.services.sagemaker.SageMakerClient;
 import org.elasticsearch.xpack.inference.services.sagemaker.SageMakerService;
 import org.elasticsearch.xpack.inference.services.sagemaker.model.SageMakerConfiguration;
@@ -492,6 +493,7 @@ public List<InferenceServiceExtension.Factory> getInferenceServiceFactories() {
             context -> new DeepSeekService(httpFactory.get(), serviceComponents.get(), context),
             context -> new LlamaService(httpFactory.get(), serviceComponents.get(), context),
             context -> new Ai21Service(httpFactory.get(), serviceComponents.get(), context),
+            context -> new OpenShiftAiService(httpFactory.get(), serviceComponents.get(), context),
             ElasticsearchInternalService::new,
             context -> new CustomService(httpFactory.get(), serviceComponents.get(), context)
         );

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/services/ServiceFields.java

@@ -14,6 +14,7 @@ public final class ServiceFields {
 
     public static final String SIMILARITY = "similarity";
     public static final String DIMENSIONS = "dimensions";
+    public static final String DIMENSIONS_SET_BY_USER = "dimensions_set_by_user";
     // Typically we use this to define the maximum tokens for the input text (text being sent to an integration)
     public static final String MAX_INPUT_TOKENS = "max_input_tokens";
     public static final String URL = "url";

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/services/openshiftai/OpenShiftAiModel.java

@@ -0,0 +1,69 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License
+ * 2.0; you may not use this file except in compliance with the Elastic License
+ * 2.0.
+ */
+
+package org.elasticsearch.xpack.inference.services.openshiftai;
+
+import org.elasticsearch.inference.ModelConfigurations;
+import org.elasticsearch.inference.ModelSecrets;
+import org.elasticsearch.inference.ServiceSettings;
+import org.elasticsearch.inference.TaskSettings;
+import org.elasticsearch.xpack.inference.external.action.ExecutableAction;
+import org.elasticsearch.xpack.inference.services.RateLimitGroupingModel;
+import org.elasticsearch.xpack.inference.services.openshiftai.action.OpenShiftAiActionVisitor;
+import org.elasticsearch.xpack.inference.services.settings.DefaultSecretSettings;
+import org.elasticsearch.xpack.inference.services.settings.RateLimitSettings;
+
+import java.util.Map;
+import java.util.Objects;
+
+/**
+ * Represents an OpenShift AI model that can be used for inference tasks.
+ * This class extends RateLimitGroupingModel to handle rate limiting based on modelId and API key.
+ */
+public abstract class OpenShiftAiModel extends RateLimitGroupingModel {
+
+    protected OpenShiftAiModel(ModelConfigurations configurations, ModelSecrets secrets) {
+        super(configurations, secrets);
+    }
+
+    protected OpenShiftAiModel(RateLimitGroupingModel model, ServiceSettings serviceSettings) {
+        super(model, serviceSettings);
+    }
+
+    protected OpenShiftAiModel(RateLimitGroupingModel model, TaskSettings taskSettings) {
+        super(model, taskSettings);
+    }
+
+    @Override
+    public RateLimitSettings rateLimitSettings() {
+        return getServiceSettings().rateLimitSettings();
+    }
+
+    @Override
+    public int rateLimitGroupingHash() {
+        return Objects.hash(getServiceSettings().uri(), getServiceSettings().modelId());
+    }
+
+    @Override
+    public OpenShiftAiServiceSettings getServiceSettings() {
+        return (OpenShiftAiServiceSettings) super.getServiceSettings();
+    }
+
+    @Override
+    public DefaultSecretSettings getSecretSettings() {
+        return (DefaultSecretSettings) super.getSecretSettings();
+    }
+
+    /**
+     * Accepts a visitor to create an executable action for this OpenShift AI model.
+     *
+     * @param creator the visitor that creates the executable action
+     * @param taskSettings the task settings to be used for the executable action
+     * @return an {@link ExecutableAction} specific to this OpenShift AI model
+     */
+    public abstract ExecutableAction accept(OpenShiftAiActionVisitor creator, Map<String, Object> taskSettings);
+}