diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index d820494c..bb5c7b91 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -9,22 +9,23 @@ jobs:
   docker:
     runs-on: ubuntu-latest
     steps:
-      -
-        name: Set up QEMU
-        uses: docker/setup-qemu-action@v2
-      -
-        name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
-      -
-        name: Login to Docker Hub
-        uses: docker/login-action@v2
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_PWD }}
       - name: Build and push
-        uses: docker/build-push-action@v3
+        uses: docker/build-push-action@v5
         with:
           tags: |
             n4z3m/dialoqbase:latest
             n4z3m/dialoqbase:${{ github.ref_name}}
           push: true
+          cache-from: type=registry,ref=n4z3m/dialoqbase:latest
+          cache-to: type=inline
diff --git a/.github/workflows/next.yml b/.github/workflows/next.yml
index 73506a7a..c08894d0 100644
--- a/.github/workflows/next.yml
+++ b/.github/workflows/next.yml
@@ -8,21 +8,22 @@ jobs:
   docker:
     runs-on: ubuntu-latest
     steps:
-      -
-        name: Set up QEMU
-        uses: docker/setup-qemu-action@v2
-      -
-        name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
-      -
-        name: Login to Docker Hub
-        uses: docker/login-action@v2
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_PWD }}
       - name: Build and push
-        uses: docker/build-push-action@v3
+        uses: docker/build-push-action@v5
         with:
           tags: |
             n4z3m/dialoqbase-next:latest
-          push: true
\ No newline at end of file
+          push: true
+          cache-from: type=registry,ref=n4z3m/dialoqbase-next:latest
+          cache-to: type=inline
\ No newline at end of file
diff --git a/app/ui/package.json b/app/ui/package.json
index f635292e..0c341779 100644
--- a/app/ui/package.json
+++ b/app/ui/package.json
@@ -1,7 +1,7 @@
 {
   "name": "app",
   "private": true,
-  "version": "1.2.1",
+  "version": "1.3.0",
   "type": "module",
   "scripts": {
     "dev": "vite",
diff --git a/app/ui/src/@types/bot.ts b/app/ui/src/@types/bot.ts
index 61ac33c2..3547bc40 100644
--- a/app/ui/src/@types/bot.ts
+++ b/app/ui/src/@types/bot.ts
@@ -13,6 +13,7 @@ export type BotSettings = {
   use_hybrid_search: boolean;
   bot_protect: boolean;
   use_rag: boolean;
+  bot_model_api_key: string;
   },
   chatModel: {
     label: string;
diff --git a/app/ui/src/components/Bot/Playground/Form.tsx b/app/ui/src/components/Bot/Playground/Form.tsx
index 14a8bac0..8dff6dfd 100644
--- a/app/ui/src/components/Bot/Playground/Form.tsx
+++ b/app/ui/src/components/Bot/Playground/Form.tsx
@@ -30,7 +30,6 @@ export const PlaygroundgForm = () => {
     stop,
   } = useSpeechRecognition();
 
-
   React.useEffect(() => {
     const defaultLanguageFromLocalStorage = localStorage.getItem(
       "defaultSpeechToTextLanguage"
diff --git a/app/ui/src/components/Bot/Settings/SettingsCard.tsx b/app/ui/src/components/Bot/Settings/SettingsCard.tsx
index 18650046..bc35b0ef 100644
--- a/app/ui/src/components/Bot/Settings/SettingsCard.tsx
+++ b/app/ui/src/components/Bot/Settings/SettingsCard.tsx
@@ -126,6 +126,7 @@ export const SettingsCard: React.FC<BotSettings> = ({
             use_hybrid_search: data.use_hybrid_search,
             bot_protect: data.bot_protect,
             use_rag: data.use_rag,
+            bot_model_api_key: data.bot_model_api_key,
           }}
           form={form}
           requiredMark={false}
@@ -304,16 +305,23 @@ export const SettingsCard: React.FC<BotSettings> = ({
 
                 <Form.Item
                   name="use_hybrid_search"
-                  label="Use Hybrid Search Retrieval (Beta)"
+                  label="Use Hybrid Search Retrieval"
                   valuePropName="checked"
                   tooltip="This will use the hybrid search retrieval method instead of the default semantic search retrieval method. Only work on playground ui."
                 >
                   <Switch />
                 </Form.Item>
+                <Form.Item
+                  name="use_rag"
+                  label="Use Retrieval Augmented Generation (RAG)"
+                  valuePropName="checked"
+                >
+                  <Switch />
+                </Form.Item>
 
                 <Form.Item
                   name="bot_protect"
-                  label="Activate Public Bot Protection (Beta)"
+                  label="Activate Public Bot Protection"
                   valuePropName="checked"
                   tooltip="This will activate the public bot protection using session to avoid misuse of the bot"
                 >
@@ -321,11 +329,15 @@ export const SettingsCard: React.FC<BotSettings> = ({
                 </Form.Item>
 
                 <Form.Item
-                  name="use_rag"
-                  label="Use Retrieval Augmented Generation (RAG)"
-                  valuePropName="checked"
+                  name="bot_model_api_key"
+                  label="Chat Model API Key"
+                  help="Enter your API key here. If you don't have one, you can leave this field blank."
+                  tooltip="Enter your API key to use your own chat model. Currently, only OpenAI API keys are supported."
                 >
-                  <Switch />
+                  <input
+                    type="password"
+                    className="mt-1 block w-full sm:text-sm focus:ring-indigo-500 focus:border-indigo-500 border-gray-300 rounded-md"
+                  />
                 </Form.Item>
               </div>
             </div>
diff --git a/app/ui/src/routes/settings/model.tsx b/app/ui/src/routes/settings/model.tsx
index 24eff6a3..05478ea8 100644
--- a/app/ui/src/routes/settings/model.tsx
+++ b/app/ui/src/routes/settings/model.tsx
@@ -75,6 +75,7 @@ export default function SettingsModelRoot() {
     const response = await api.post("/admin/models", {
       ...values,
       url: fetchUrlForm.getFieldValue("url"),
+      api_key: fetchUrlForm.getFieldValue("api_key"),
     });
     return response.data;
   };
@@ -311,6 +312,14 @@ export default function SettingsModelRoot() {
                   />
                 </Form.Item>
 
+                <Form.Item name="api_key" label="API Key (Optional)">
+                  <input
+                    type="text"
+                    className="border border-gray-200 rounded-md px-3 py-2 w-full"
+                    placeholder="API Key (Optional)"
+                  />
+                </Form.Item>
+
                 <button
                   type="submit"
                   disabled={isSaveModel}
diff --git a/package.json b/package.json
index fb2f2f4e..70487a92 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "dialoqbase",
-  "version": "1.2.1",
+  "version": "1.3.0",
   "description": "Create chatbots with ease",
   "scripts": {
     "ui:dev": "pnpm run --filter ui dev",
diff --git a/server/prisma/migrations/20231102171748_q_17/migration.sql b/server/prisma/migrations/20231102171748_q_17/migration.sql
new file mode 100644
index 00000000..24fdacaa
--- /dev/null
+++ b/server/prisma/migrations/20231102171748_q_17/migration.sql
@@ -0,0 +1,2 @@
+-- AlterTable
+ALTER TABLE "Bot" ADD COLUMN     "bot_model_api_key" TEXT;
diff --git a/server/prisma/schema.prisma b/server/prisma/schema.prisma
index c3505377..c73097df 100644
--- a/server/prisma/schema.prisma
+++ b/server/prisma/schema.prisma
@@ -34,6 +34,7 @@ model Bot {
   use_rag                     Boolean          @default(false)
   bot_protect                 Boolean          @default(false)
   bot_api_key                 String?
+  bot_model_api_key           String?
   options                     Json?            @default("{}") @db.Json
   BotAppearance               BotAppearance[]
   document                    BotDocument[]
diff --git a/server/src/integration/handlers/discord.handler.ts b/server/src/integration/handlers/discord.handler.ts
index a57f23d6..69d80417 100644
--- a/server/src/integration/handlers/discord.handler.ts
+++ b/server/src/integration/handlers/discord.handler.ts
@@ -69,7 +69,14 @@ let history = chat_history
       }
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
 
     const model = chatModelProvider(bot.provider, bot.model, temperature, {
       ...botConfig,
diff --git a/server/src/integration/handlers/telegram.handler.ts b/server/src/integration/handlers/telegram.handler.ts
index 3c9d91ab..0c40595d 100644
--- a/server/src/integration/handlers/telegram.handler.ts
+++ b/server/src/integration/handlers/telegram.handler.ts
@@ -67,7 +67,15 @@ export const telegramBotHandler = async (
       return "Unable to find model";
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
+
     const model = chatModelProvider(bot.provider, bot.model, temperature, {
       ...botConfig,
     });
diff --git a/server/src/integration/handlers/whatsapp.handler.ts b/server/src/integration/handlers/whatsapp.handler.ts
index fdf88a8b..8a6bcbb7 100644
--- a/server/src/integration/handlers/whatsapp.handler.ts
+++ b/server/src/integration/handlers/whatsapp.handler.ts
@@ -9,7 +9,7 @@ export const whatsappBotHandler = async (
   bot_id: string,
   hash: string,
   from: string,
-  message: string,
+  message: string
 ) => {
   try {
     await prisma.$connect();
@@ -45,9 +45,11 @@ export const whatsappBotHandler = async (
       chat_history.splice(0, chat_history.length - 10);
     }
 
-    let history = chat_history.map((chat) => {
-      return `Human: ${chat.human}\nAssistant: ${chat.bot}`;
-    }).join("\n");
+    let history = chat_history
+      .map((chat) => {
+        return `Human: ${chat.human}\nAssistant: ${chat.bot}`;
+      })
+      .join("\n");
 
     const temperature = bot.temperature;
 
@@ -59,7 +61,7 @@ export const whatsappBotHandler = async (
       {
         botId: bot.id,
         sourceId: null,
-      },
+      }
     );
 
     const modelinfo = await prisma.dialoqbaseModels.findFirst({
@@ -74,16 +76,18 @@ export const whatsappBotHandler = async (
       return "Unable to find model";
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
 
-    const model = chatModelProvider(
-      bot.provider,
-      bot.model,
-      temperature,
-      {
-        ...botConfig,
-      },
-    );
+    const model = chatModelProvider(bot.provider, bot.model, temperature, {
+      ...botConfig,
+    });
 
     const chain = ConversationalRetrievalQAChain.fromLLM(
       model,
@@ -92,7 +96,7 @@ export const whatsappBotHandler = async (
         qaTemplate: bot.qaPrompt,
         questionGeneratorTemplate: bot.questionGeneratorPrompt,
         returnSourceDocuments: true,
-      },
+      }
     );
 
     const response = await chain.call({
@@ -121,12 +125,8 @@ export const whatsappBotHandler = async (
   }
 };
 
-export const clearHistoryWhatsapp = async (
-    bot_id: string,
-  from: string,
-) => {
+export const clearHistoryWhatsapp = async (bot_id: string, from: string) => {
   try {
-
     await prisma.$connect();
 
     const bot = await prisma.bot.findFirst({
diff --git a/server/src/routes/api/v1/admin/handlers/model.handler.ts b/server/src/routes/api/v1/admin/handlers/model.handler.ts
index 36fcda8d..1082d74d 100644
--- a/server/src/routes/api/v1/admin/handlers/model.handler.ts
+++ b/server/src/routes/api/v1/admin/handlers/model.handler.ts
@@ -8,9 +8,16 @@ import {
 } from "./type";
 import axios from "axios";
 
-const _getModelFromUrl = async (url: string) => {
+const _getModelFromUrl = async (url: string, apiKey?: string) => {
   try {
-    const response = await axios.get(`${url}/models`);
+    const response = await axios.get(`${url}/models`, {
+      headers: {
+        "HTTP-Referer":
+          process.env.LOCAL_REFER_URL || "https://dialoqbase.n4ze3m.com/",
+        "X-Title": process.env.LOCAL_TITLE || "Dialoqbase",
+        Authorization: apiKey && `Bearer ${apiKey}`,
+      },
+    });
     return response.data as {
       type: string;
       data: {
@@ -58,7 +65,7 @@ export const fetchModelFromInputedUrlHandler = async (
   reply: FastifyReply
 ) => {
   try {
-    const { url } = request.body;
+    const { url, api_key } = request.body;
     const user = request.user;
     if (!user.is_admin) {
       return reply.status(403).send({
@@ -66,7 +73,7 @@ export const fetchModelFromInputedUrlHandler = async (
       });
     }
 
-    const model = await _getModelFromUrl(url);
+    const model = await _getModelFromUrl(url, api_key);
 
     if (!model) {
       return reply.status(404).send({
@@ -100,7 +107,7 @@ export const saveModelFromInputedUrlHandler = async (
       });
     }
 
-    const { url, model_id, name, stream_available } = request.body;
+    const { url, api_key, model_id, name, stream_available } = request.body;
 
     const modelExist = await prisma.dialoqbaseModels.findFirst({
       where: {
@@ -125,6 +132,7 @@ export const saveModelFromInputedUrlHandler = async (
         model_provider: "local",
         config: {
           baseURL: url,
+          apiKey: api_key,
         },
       },
     });
diff --git a/server/src/routes/api/v1/admin/handlers/type.ts b/server/src/routes/api/v1/admin/handlers/type.ts
index 15d19ccf..211f70c2 100644
--- a/server/src/routes/api/v1/admin/handlers/type.ts
+++ b/server/src/routes/api/v1/admin/handlers/type.ts
@@ -25,6 +25,7 @@ export type RegisterUserbyAdminRequestBody = {
 export type FetchModelFromInputedUrlRequest = {
   Body: {
     url: string;
+    api_key?: string;
   };
 };
 export type SaveModelFromInputedUrlRequest = {
@@ -33,6 +34,7 @@ export type SaveModelFromInputedUrlRequest = {
     model_id: string;
     name: string;
     stream_available: boolean;
+    api_key?: string;
   };
 };
 
diff --git a/server/src/routes/api/v1/bot/handlers/types.ts b/server/src/routes/api/v1/bot/handlers/types.ts
index 90722df0..a202aec3 100644
--- a/server/src/routes/api/v1/bot/handlers/types.ts
+++ b/server/src/routes/api/v1/bot/handlers/types.ts
@@ -66,5 +66,6 @@ export interface UpdateBotById {
     use_hybrid_search: boolean;
     bot_protect: boolean;
     use_rag: boolean;
+    bot_model_api_key: string
   };
 }
diff --git a/server/src/routes/api/v1/bot/playground/handlers/post.handler.ts b/server/src/routes/api/v1/bot/playground/handlers/post.handler.ts
index 92fb9717..13a9c78b 100644
--- a/server/src/routes/api/v1/bot/playground/handlers/post.handler.ts
+++ b/server/src/routes/api/v1/bot/playground/handlers/post.handler.ts
@@ -13,35 +13,12 @@ import {
 import { StringOutputParser } from "langchain/schema/output_parser";
 import { PromptTemplate } from "langchain/prompts";
 import { Document } from "langchain/document";
-
-type ChatMessage = {
-  type: "human" | "ai" | "other";
-  text: string;
-};
-
-type ConversationalRetrievalQAChainInput = {
-  question: string;
-  chat_history: ChatMessage[];
-};
-
-const formatChatHistory = (history: ChatMessage[]): string => {
-  return history
-    .map((chatMessage: ChatMessage) => {
-      if (chatMessage.type === "human") {
-        return `Human: ${chatMessage.text}`;
-      } else if (chatMessage.type === "ai") {
-        return `Assistant: ${chatMessage.text}`;
-      } else {
-        return `${chatMessage.text}`;
-      }
-    })
-    .join("\n");
-};
-
-const combineDocumentsFn = (docs: Document[], separator = "\n\n") => {
-  const serializedDocs = docs.map((doc) => doc.pageContent);
-  return serializedDocs.join(separator);
-};
+import {
+  ChatMessage,
+  ConversationalRetrievalQAChainInput,
+  combineDocumentsFn,
+  formatChatHistory,
+} from "../../../../../../utils/rag";
 
 export const chatRequestHandler = async (
   request: FastifyRequest<ChatRequestBody>,
@@ -149,7 +126,15 @@ export const chatRequestHandler = async (
       };
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
+
 
     const model = chatModelProvider(bot.provider, bot.model, temperature, {
       ...botConfig,
@@ -289,14 +274,9 @@ export const chatRequestHandler = async (
         },
       });
 
-      const response = await chain.invoke({
-        question: sanitizedQuestion,
-        chat_history: history as ChatMessage[],
-      });
-
       return {
         bot: {
-          text: response.content,
+          text: botResponse.content,
           sourceDocuments: documents,
         },
         history: [
@@ -307,7 +287,7 @@ export const chatRequestHandler = async (
           },
           {
             type: "ai",
-            text: response.content,
+            text: botResponse.content,
           },
         ],
       };
@@ -455,7 +435,15 @@ export const chatRequestStreamHandler = async (
 
       return reply.raw.end();
     }
-    const botConfig = (modelinfo.config as {}) || {};
+
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
 
     let response: any = null;
     const streamedModel = chatModelProvider(
diff --git a/server/src/routes/api/v1/bot/schema/index.ts b/server/src/routes/api/v1/bot/schema/index.ts
index b642243b..37631fff 100644
--- a/server/src/routes/api/v1/bot/schema/index.ts
+++ b/server/src/routes/api/v1/bot/schema/index.ts
@@ -125,6 +125,9 @@ export const updateBotByIdSchema: FastifySchema = {
       use_rag: {
         type: "boolean",
       },
+      bot_model_api_key: {
+        type: "string",
+      }
     },
   },
 };
diff --git a/server/src/routes/api/v1/user/handlers/post.handler.ts b/server/src/routes/api/v1/user/handlers/post.handler.ts
index 6f2578d0..ebabd83d 100644
--- a/server/src/routes/api/v1/user/handlers/post.handler.ts
+++ b/server/src/routes/api/v1/user/handlers/post.handler.ts
@@ -224,6 +224,7 @@ export const registerUserHandler = async (
       to: "/",
     });
   } catch (e) {
+    console.log(e);
     return reply.status(500).send({
       message: "Error registering user",
     });
diff --git a/server/src/routes/bot/handlers/post.handler.ts b/server/src/routes/bot/handlers/post.handler.ts
index 99f20854..79816af1 100644
--- a/server/src/routes/bot/handlers/post.handler.ts
+++ b/server/src/routes/bot/handlers/post.handler.ts
@@ -4,6 +4,21 @@ import { DialoqbaseVectorStore } from "../../../utils/store";
 import { ConversationalRetrievalQAChain } from "langchain/chains";
 import { embeddings } from "../../../utils/embeddings";
 import { chatModelProvider } from "../../../utils/models";
+import { BaseRetriever } from "langchain/schema/retriever";
+import { DialoqbaseHybridRetrival } from "../../../utils/hybrid";
+import { Document } from "langchain/document";
+import { PromptTemplate } from "langchain/prompts";
+import {
+  RunnablePassthrough,
+  RunnableSequence,
+} from "langchain/schema/runnable";
+import { StringOutputParser } from "langchain/schema/output_parser";
+import {
+  ChatMessage,
+  ConversationalRetrievalQAChainInput,
+  combineDocumentsFn,
+  formatChatHistory,
+} from "../../../utils/rag";
 
 export const chatRequestHandler = async (
   request: FastifyRequest<ChatRequestBody>,
@@ -67,14 +82,42 @@ export const chatRequestHandler = async (
 
     const sanitizedQuestion = message.trim().replaceAll("\n", " ");
     const embeddingModel = embeddings(bot.embedding);
-
-    const vectorstore = await DialoqbaseVectorStore.fromExistingIndex(
-      embeddingModel,
-      {
+    let retriever: BaseRetriever;
+    let resolveWithDocuments: (value: Document[]) => void;
+    const documentPromise = new Promise<Document[]>((resolve) => {
+      resolveWithDocuments = resolve;
+    });
+    if (bot.use_hybrid_search) {
+      retriever = new DialoqbaseHybridRetrival(embeddingModel, {
         botId: bot.id,
         sourceId: null,
-      }
-    );
+        callbacks: [
+          {
+            handleRetrieverEnd(documents) {
+              resolveWithDocuments(documents);
+            },
+          },
+        ],
+      });
+    } else {
+      const vectorstore = await DialoqbaseVectorStore.fromExistingIndex(
+        embeddingModel,
+        {
+          botId: bot.id,
+          sourceId: null,
+        }
+      );
+
+      retriever = vectorstore.asRetriever({
+        callbacks: [
+          {
+            handleRetrieverEnd(documents) {
+              resolveWithDocuments(documents);
+            },
+          },
+        ],
+      });
+    }
 
     const modelinfo = await prisma.dialoqbaseModels.findFirst({
       where: {
@@ -104,67 +147,138 @@ export const chatRequestHandler = async (
       };
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
 
     const model = chatModelProvider(bot.provider, bot.model, temperature, {
       ...botConfig,
     });
 
-    const chain = ConversationalRetrievalQAChain.fromLLM(
-      model,
-      vectorstore.asRetriever(),
-      {
+    if (!bot.use_rag) {
+      const chain = ConversationalRetrievalQAChain.fromLLM(model, retriever, {
         qaTemplate: bot.qaPrompt,
         questionGeneratorTemplate: bot.questionGeneratorPrompt,
         returnSourceDocuments: true,
-      }
-    );
-
-    const chat_history = history
-      .map((chatMessage: any) => {
-        if (chatMessage.type === "human") {
-          return `Human: ${chatMessage.text}`;
-        } else if (chatMessage.type === "ai") {
-          return `Assistant: ${chatMessage.text}`;
-        } else {
-          return `${chatMessage.text}`;
-        }
-      })
-      .join("\n");
+      });
 
-    const response = await chain.call({
-      question: sanitizedQuestion,
-      chat_history: chat_history,
-    });
+      const chat_history = history
+        .map((chatMessage: any) => {
+          if (chatMessage.type === "human") {
+            return `Human: ${chatMessage.text}`;
+          } else if (chatMessage.type === "ai") {
+            return `Assistant: ${chatMessage.text}`;
+          } else {
+            return `${chatMessage.text}`;
+          }
+        })
+        .join("\n");
+
+      const response = await chain.call({
+        question: sanitizedQuestion,
+        chat_history: chat_history,
+      });
 
-    await prisma.botWebHistory.create({
-      data: {
-        chat_id: history_id,
-        bot_id: bot.id,
-        bot: response.text,
-        human: message,
-        metadata: {
-          ip: request?.ip,
-          user_agent: request?.headers["user-agent"],
+      await prisma.botWebHistory.create({
+        data: {
+          chat_id: history_id,
+          bot_id: bot.id,
+          bot: response.text,
+          human: message,
+          metadata: {
+            ip: request?.ip,
+            user_agent: request?.headers["user-agent"],
+          },
+          sources: response?.sources,
         },
-        sources: response?.sources,
-      },
-    });
+      });
 
-    return {
-      bot: response,
-      history: [
-        ...history,
+      return {
+        bot: response,
+        history: [
+          ...history,
+          {
+            type: "human",
+            text: message,
+          },
+          {
+            type: "ai",
+            text: response.text,
+          },
+        ],
+      };
+    } else {
+      const standaloneQuestionChain = RunnableSequence.from([
         {
-          type: "human",
-          text: message,
+          question: (input: ConversationalRetrievalQAChainInput) =>
+            input.question,
+          chat_history: (input: ConversationalRetrievalQAChainInput) =>
+            formatChatHistory(input.chat_history),
         },
+        PromptTemplate.fromTemplate(bot.questionGeneratorPrompt),
+        model,
+        new StringOutputParser(),
+      ]);
+
+      //@ts-ignore
+      const answerChain = RunnableSequence.from([
         {
-          type: "ai",
-          text: response.text,
+          context: retriever.pipe(combineDocumentsFn),
+          question: new RunnablePassthrough(),
         },
-      ],
-    };
+        PromptTemplate.fromTemplate(bot.qaPrompt),
+        model,
+      ]);
+
+      const chain = standaloneQuestionChain.pipe(answerChain);
+      const botResponse = await chain.invoke({
+        question: sanitizedQuestion,
+        chat_history: history as ChatMessage[],
+      });
+
+      const documents = await documentPromise;
+
+      await prisma.botWebHistory.create({
+        data: {
+          chat_id: history_id,
+          bot_id: bot.id,
+          bot: botResponse.content,
+          human: message,
+          metadata: {
+            ip: request?.ip,
+            user_agent: request?.headers["user-agent"],
+          },
+          sources: documents.map((doc) => {
+            return {
+              ...doc,
+            };
+          }),
+        },
+      });
+
+      return {
+        bot: {
+          text: botResponse.content,
+          sourceDocuments: documents,
+        },
+        history: [
+          ...history,
+          {
+            type: "human",
+            text: message,
+          },
+          {
+            type: "ai",
+            text: botResponse.content,
+          },
+        ],
+      };
+    }
   } catch (e) {
     return {
       bot: {
@@ -274,13 +388,42 @@ export const chatRequestStreamHandler = async (
     const sanitizedQuestion = message.trim().replaceAll("\n", " ");
     const embeddingModel = embeddings(bot.embedding);
 
-    const vectorstore = await DialoqbaseVectorStore.fromExistingIndex(
-      embeddingModel,
-      {
+    let retriever: BaseRetriever;
+    let resolveWithDocuments: (value: Document[]) => void;
+    const documentPromise = new Promise<Document[]>((resolve) => {
+      resolveWithDocuments = resolve;
+    });
+    if (bot.use_hybrid_search) {
+      retriever = new DialoqbaseHybridRetrival(embeddingModel, {
         botId: bot.id,
         sourceId: null,
-      }
-    );
+        callbacks: [
+          {
+            handleRetrieverEnd(documents) {
+              resolveWithDocuments(documents);
+            },
+          },
+        ],
+      });
+    } else {
+      const vectorstore = await DialoqbaseVectorStore.fromExistingIndex(
+        embeddingModel,
+        {
+          botId: bot.id,
+          sourceId: null,
+        }
+      );
+
+      retriever = vectorstore.asRetriever({
+        callbacks: [
+          {
+            handleRetrieverEnd(documents) {
+              resolveWithDocuments(documents);
+            },
+          },
+        ],
+      });
+    }
 
     let response: any = null;
 
@@ -325,7 +468,14 @@ export const chatRequestStreamHandler = async (
       return reply.raw.end();
     }
 
-    const botConfig = (modelinfo.config as {}) || {};
+    const botConfig: any = (modelinfo.config as {}) || {};
+    if (bot.provider.toLowerCase() === "openai") {
+      if (bot.bot_model_api_key && bot.bot_model_api_key.trim() !== "") {
+        botConfig.configuration = {
+          apiKey: bot.bot_model_api_key,
+        };
+      }
+    }
 
     const streamedModel = chatModelProvider(
       bot.provider,
@@ -364,73 +514,146 @@ export const chatRequestStreamHandler = async (
         ...botConfig,
       }
     );
-
-    const chain = ConversationalRetrievalQAChain.fromLLM(
-      streamedModel,
-      vectorstore.asRetriever(),
-      {
-        qaTemplate: bot.qaPrompt,
-        questionGeneratorTemplate: bot.questionGeneratorPrompt,
-        returnSourceDocuments: true,
-        questionGeneratorChainOptions: {
-          llm: nonStreamingModel,
-        },
-      }
-    );
-
-    const chat_history = history
-      .map((chatMessage: any) => {
-        if (chatMessage.type === "human") {
-          return `Human: ${chatMessage.text}`;
-        } else if (chatMessage.type === "ai") {
-          return `Assistant: ${chatMessage.text}`;
-        } else {
-          return `${chatMessage.text}`;
+    if (!bot.use_rag) {
+      const chain = ConversationalRetrievalQAChain.fromLLM(
+        streamedModel,
+        retriever,
+        {
+          qaTemplate: bot.qaPrompt,
+          questionGeneratorTemplate: bot.questionGeneratorPrompt,
+          returnSourceDocuments: true,
+          questionGeneratorChainOptions: {
+            llm: nonStreamingModel,
+          },
         }
-      })
-      .join("\n");
+      );
+
+      const chat_history = history
+        .map((chatMessage: any) => {
+          if (chatMessage.type === "human") {
+            return `Human: ${chatMessage.text}`;
+          } else if (chatMessage.type === "ai") {
+            return `Assistant: ${chatMessage.text}`;
+          } else {
+            return `${chatMessage.text}`;
+          }
+        })
+        .join("\n");
+
+      console.log("Waiting for response...");
+
+      response = await chain.call({
+        question: sanitizedQuestion,
+        chat_history: chat_history,
+      });
 
-    console.log("Waiting for response...");
+      await prisma.botWebHistory.create({
+        data: {
+          chat_id: history_id,
+          bot_id: bot.id,
+          bot: response.text,
+          human: message,
+          metadata: {
+            ip: request?.ip,
+            user_agent: request?.headers["user-agent"],
+          },
+          sources: response?.sources || response?.sourceDocuments || [],
+        },
+      });
 
-    response = await chain.call({
-      question: sanitizedQuestion,
-      chat_history: chat_history,
-    });
+      reply.sse({
+        event: "result",
+        id: "",
+        data: JSON.stringify({
+          bot: response,
+          history: [
+            ...history,
+            {
+              type: "human",
+              text: message,
+            },
+            {
+              type: "ai",
+              text: response.text,
+            },
+          ],
+        }),
+      });
+      await nextTick();
+      return reply.raw.end();
+    } else {
+      const standaloneQuestionChain = RunnableSequence.from([
+        {
+          question: (input: ConversationalRetrievalQAChainInput) =>
+            input.question,
+          chat_history: (input: ConversationalRetrievalQAChainInput) =>
+            formatChatHistory(input.chat_history),
+        },
+        PromptTemplate.fromTemplate(bot.questionGeneratorPrompt),
+        nonStreamingModel,
+        new StringOutputParser(),
+      ]);
 
-    await prisma.botWebHistory.create({
-      data: {
-        chat_id: history_id,
-        bot_id: bot.id,
-        bot: response.text,
-        human: message,
-        metadata: {
-          ip: request?.ip,
-          user_agent: request?.headers["user-agent"],
+      //@ts-ignore
+      const answerChain = RunnableSequence.from([
+        {
+          context: retriever.pipe(combineDocumentsFn),
+          question: new RunnablePassthrough(),
         },
-        sources: response?.sources || response?.sourceDocuments || [],
-      },
-    });
+        PromptTemplate.fromTemplate(bot.qaPrompt),
+        streamedModel,
+      ]);
+
+      const chain = standaloneQuestionChain.pipe(answerChain);
+      response = await chain.invoke({
+        question: sanitizedQuestion,
+        chat_history: history as ChatMessage[],
+      });
 
-    reply.sse({
-      event: "result",
-      id: "",
-      data: JSON.stringify({
-        bot: response,
-        history: [
-          ...history,
-          {
-            type: "human",
-            text: message,
+      const documents = await documentPromise;
+
+      await prisma.botWebHistory.create({
+        data: {
+          chat_id: history_id,
+          bot_id: bot.id,
+          bot: response.text,
+          human: message,
+          metadata: {
+            ip: request?.ip,
+            user_agent: request?.headers["user-agent"],
           },
-          {
-            type: "ai",
-            text: response.text,
+          sources: documents.map((doc) => {
+            return {
+              ...doc,
+            };
+          }),
+        },
+      });
+
+      reply.sse({
+        event: "result",
+        id: "",
+        data: JSON.stringify({
+          bot: {
+            text: response.content,
+            sourceDocuments: documents,
           },
-        ],
-      }),
-    });
-    await nextTick();
-    return reply.raw.end();
+          history: [
+            ...history,
+            {
+              type: "human",
+              text: message,
+            },
+            {
+              type: "ai",
+              text: response.text,
+            },
+          ],
+        }),
+      });
+      await nextTick();
+      return reply.raw.end();
+    }
   } catch (e) {
     console.log(e);
     reply.raw.setHeader("Content-Type", "text/event-stream");
diff --git a/server/src/utils/models.ts b/server/src/utils/models.ts
index b4182370..fc264616 100644
--- a/server/src/utils/models.ts
+++ b/server/src/utils/models.ts
@@ -16,14 +16,17 @@ export const chatModelProvider = (
   console.log("provider", provider);
   console.log("modelName", modelName);
 
-
   switch (provider.toLowerCase()) {
     case "openai":
-      console.log("using openai");
+      console.log("using openai", otherFields);
       return new ChatOpenAI({
         modelName: modelName,
         temperature: temperature,
         ...otherFields,
+        configuration: {
+          ...otherFields.configuration,
+          baseURL: process.env.OPENAI_API_URL,
+        },
       });
     case "anthropic":
       console.log("using anthropic");
@@ -59,6 +62,9 @@ export const chatModelProvider = (
         modelName: modelName,
         temperature: temperature,
         ...otherFields,
+        configuration: {
+          baseURL: process.env.OPENAI_API_URL,
+        },
       });
     case "local":
       console.log("using local");
@@ -68,6 +74,11 @@ export const chatModelProvider = (
         ...otherFields,
         configuration: {
           baseURL: otherFields.baseURL,
+          apiKey: otherFields.apiKey || process.env.OPENAI_API_KEY,
+          defaultHeaders: {
+            "HTTP-Referer": process.env.LOCAL_REFER_URL || "https://dialoqbase.n4ze3m.com/",
+            "X-Title": process.env.LOCAL_TITLE || "Dialoqbase",
+          },
         },
       });
     default:
@@ -80,28 +91,6 @@ export const chatModelProvider = (
   }
 };
 
-export const huggingfaceModels: {
-  [key: string]: string;
-} = {
-  "falcon-7b-instruct-inference": "tiiuae/falcon-7b-instruct",
-};
-
-export const fireworksModels: {
-  [key: string]: string;
-} = {
-  "llama-v2-7b-chat": "accounts/fireworks/models/llama-v2-7b-chat",
-  "llama-v2-13b-chat": "accounts/fireworks/models/llama-v2-13b-chat",
-  "llama-v2-70b-chat": "accounts/fireworks/models/llama-v2-70b-chat",
-  "llama-v2-7b-chat-w8a16": "accounts/fireworks/models/llama-v2-7b-chat-w8a16",
-  "llama-v2-13b-chat-w8a16":
-    "accounts/fireworks/models/llama-v2-13b-chat-w8a16",
-  "llama-v2-13b-code-instruct":
-    "accounts/fireworks/models/llama-v2-13b-code-instruct",
-  "llama-v2-34b-code-instruct-w8a16":
-    "accounts/fireworks/models/llama-v2-34b-code-instruct-w8a16",
-  "mistral-7b-instruct-4k": "accounts/fireworks/models/mistral-7b-instruct-4k",
-};
-
 export const streamingSupportedModels = [
   "gpt-3.5-turbo",
   "gpt-3.5-turbo-16k",
diff --git a/server/src/utils/rag.ts b/server/src/utils/rag.ts
new file mode 100644
index 00000000..069fd323
--- /dev/null
+++ b/server/src/utils/rag.ts
@@ -0,0 +1,30 @@
+import { Document } from "langchain/document";
+
+export type ChatMessage = {
+  type: "human" | "ai" | "other";
+  text: string;
+};
+
+export type ConversationalRetrievalQAChainInput = {
+  question: string;
+  chat_history: ChatMessage[];
+};
+
+export const formatChatHistory = (history: ChatMessage[]): string => {
+  return history
+    .map((chatMessage: ChatMessage) => {
+      if (chatMessage.type === "human") {
+        return `Human: ${chatMessage.text}`;
+      } else if (chatMessage.type === "ai") {
+        return `Assistant: ${chatMessage.text}`;
+      } else {
+        return `${chatMessage.text}`;
+      }
+    })
+    .join("\n");
+};
+
+export const combineDocumentsFn = (docs: Document[], separator = "\n\n") => {
+  const serializedDocs = docs.map((doc) => doc.pageContent);
+  return serializedDocs.join(separator);
+};