diff --git a/config/settings.yml b/config/settings.yml
index 4fcf3a8b..71e78e96 100644
--- a/config/settings.yml
+++ b/config/settings.yml
@@ -84,3 +84,6 @@ plugins:
     choices:
      - opennsfw2
      - nsfw_detector
+
+  ai_openai_api_key:
+    default: ""
diff --git a/lib/modules/nsfw/nsfw_classification.rb b/lib/modules/nsfw/nsfw_classification.rb
index a63dd992..f3d5967f 100644
--- a/lib/modules/nsfw/nsfw_classification.rb
+++ b/lib/modules/nsfw/nsfw_classification.rb
@@ -52,7 +52,7 @@ module DiscourseAI
         upload_url = Discourse.store.cdn_url(upload.url)
         upload_url = "#{Discourse.base_url_no_prefix}#{upload_url}" if upload_url.starts_with?("/")
 
-        DiscourseAI::InferenceManager.perform!(
+        DiscourseAI::Inference::DiscourseClassifier.perform!(
           "#{SiteSetting.ai_nsfw_inference_service_api_endpoint}/api/v1/classify",
           model,
           upload_url,
diff --git a/lib/modules/sentiment/sentiment_classification.rb b/lib/modules/sentiment/sentiment_classification.rb
index 76a2ba3e..633563dd 100644
--- a/lib/modules/sentiment/sentiment_classification.rb
+++ b/lib/modules/sentiment/sentiment_classification.rb
@@ -39,7 +39,7 @@ module DiscourseAI
       private
 
       def request_with(model, content)
-        ::DiscourseAI::InferenceManager.perform!(
+        ::DiscourseAI::Inference::DiscourseClassifier.perform!(
           "#{SiteSetting.ai_sentiment_inference_service_api_endpoint}/api/v1/classify",
           model,
           content,
diff --git a/lib/modules/toxicity/toxicity_classification.rb b/lib/modules/toxicity/toxicity_classification.rb
index bf4e3679..815b9269 100644
--- a/lib/modules/toxicity/toxicity_classification.rb
+++ b/lib/modules/toxicity/toxicity_classification.rb
@@ -42,7 +42,7 @@ module DiscourseAI
 
       def request(target_to_classify)
         data =
-          ::DiscourseAI::InferenceManager.perform!(
+          ::DiscourseAI::Inference::DiscourseClassifier.perform!(
             "#{SiteSetting.ai_toxicity_inference_service_api_endpoint}/api/v1/classify",
             SiteSetting.ai_toxicity_inference_service_api_model,
             content_of(target_to_classify),
diff --git a/lib/shared/inference/discourse_classifier.rb b/lib/shared/inference/discourse_classifier.rb
new file mode 100644
index 00000000..601c3245
--- /dev/null
+++ b/lib/shared/inference/discourse_classifier.rb
@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module ::DiscourseAI
+  module Inference
+    class DiscourseClassifier
+      def self.perform!(endpoint, model, content, api_key)
+        headers = { "Referer" => Discourse.base_url, "Content-Type" => "application/json" }
+
+        headers["X-API-KEY"] = api_key if api_key.present?
+
+        response = Faraday.post(endpoint, { model: model, content: content }.to_json, headers)
+
+        raise Net::HTTPBadResponse unless response.status == 200
+
+        JSON.parse(response.body, symbolize_names: true)
+      end
+    end
+  end
+end
diff --git a/lib/shared/inference/discourse_reranker.rb b/lib/shared/inference/discourse_reranker.rb
new file mode 100644
index 00000000..b50953f5
--- /dev/null
+++ b/lib/shared/inference/discourse_reranker.rb
@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module ::DiscourseAI
+  module Inference
+    class DiscourseReranker
+      def self.perform!(endpoint, model, content, candidates, api_key)
+        headers = { "Referer" => Discourse.base_url, "Content-Type" => "application/json" }
+
+        headers["X-API-KEY"] = api_key if api_key.present?
+
+        response =
+          Faraday.post(
+            endpoint,
+            { model: model, content: content, candidates: candidates }.to_json,
+            headers,
+          )
+
+        raise Net::HTTPBadResponse unless response.status == 200
+
+        JSON.parse(response.body, symbolize_names: true)
+      end
+    end
+  end
+end
diff --git a/lib/shared/inference/openai_completions.rb b/lib/shared/inference/openai_completions.rb
new file mode 100644
index 00000000..49364e5f
--- /dev/null
+++ b/lib/shared/inference/openai_completions.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module ::DiscourseAI
+  module Inference
+    class OpenAICompletions
+      def self.perform!(model, content, api_key)
+        headers = {
+          "Authorization" => "Bearer #{SiteSetting.ai_openai_api_key}",
+          "Content-Type" => "application/json",
+        }
+
+        model ||= "gpt-3.5-turbo"
+
+        response =
+          Faraday.post(
+            "https://api.openai.com/v1/chat/completions",
+            { model: model, messages: content }.to_json,
+            headers,
+          )
+
+        raise Net::HTTPBadResponse unless response.status == 200
+
+        JSON.parse(response.body, symbolize_names: true)
+      end
+    end
+  end
+end
diff --git a/lib/shared/inference/openai_embeddings.rb b/lib/shared/inference/openai_embeddings.rb
new file mode 100644
index 00000000..116bc62e
--- /dev/null
+++ b/lib/shared/inference/openai_embeddings.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module ::DiscourseAI
+  module Inference
+    class OpenAIEmbeddings
+      def self.perform!(content, model = nil)
+        headers = {
+          "Authorization" => "Bearer #{SiteSetting.ai_openai_api_key}",
+          "Content-Type" => "application/json",
+        }
+
+        model ||= "text-embedding-ada-002"
+
+        response =
+          Faraday.post(
+            "https://api.openai.com/v1/embeddings",
+            { model: model, input: content }.to_json,
+            headers,
+          )
+
+        raise Net::HTTPBadResponse unless response.status == 200
+
+        JSON.parse(response.body, symbolize_names: true)
+      end
+    end
+  end
+end
diff --git a/lib/shared/inference_manager.rb b/lib/shared/inference_manager.rb
deleted file mode 100644
index 6f7b12ee..00000000
--- a/lib/shared/inference_manager.rb
+++ /dev/null
@@ -1,17 +0,0 @@
-# frozen_string_literal: true
-
-module ::DiscourseAI
-  class InferenceManager
-    def self.perform!(endpoint, model, content, api_key)
-      headers = { "Referer" => Discourse.base_url, "Content-Type" => "application/json" }
-
-      headers["X-API-KEY"] = api_key if api_key.present?
-
-      response = Faraday.post(endpoint, { model: model, content: content }.to_json, headers)
-
-      raise Net::HTTPBadResponse unless response.status == 200
-
-      JSON.parse(response.body, symbolize_names: true)
-    end
-  end
-end
diff --git a/lib/tasks/modules/toxicity/calibration.rake b/lib/tasks/modules/toxicity/calibration.rake
index b9fe6bf3..50cf3f1c 100644
--- a/lib/tasks/modules/toxicity/calibration.rake
+++ b/lib/tasks/modules/toxicity/calibration.rake
@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 def classify(content)
-  ::DiscourseAI::InferenceManager.perform!(
+  ::DiscourseAI::Inference::DiscourseClassifier.perform!(
     "#{SiteSetting.ai_toxicity_inference_service_api_endpoint}/api/v1/classify",
     SiteSetting.ai_toxicity_inference_service_api_model,
     content,
diff --git a/plugin.rb b/plugin.rb
index 6f0e3549..7181712e 100644
--- a/plugin.rb
+++ b/plugin.rb
@@ -16,7 +16,11 @@ after_initialize do
     PLUGIN_NAME = "discourse-ai"
   end
 
-  require_relative "lib/shared/inference_manager"
+  require_relative "lib/shared/inference/discourse_classifier"
+  require_relative "lib/shared/inference/discourse_reranker"
+  require_relative "lib/shared/inference/openai_completions"
+  require_relative "lib/shared/inference/openai_embeddings"
+
   require_relative "lib/shared/classificator"
   require_relative "lib/shared/post_classificator"
   require_relative "lib/shared/chat_message_classificator"