fix: prefer using embedding model from vector store (#1708)

.changeset/tame-chairs-shake.md

+---
+"llamaindex": patch
+"@llamaindex/deepseek": patch
+"@llamaindex/fireworks": patch
+"@llamaindex/together": patch
+"@llamaindex/jinaai": patch
+---
+
+fix: prefer using embedding model from vector store

apps/next/src/content/docs/llamaindex/modules/embeddings/available_embeddings/jinaai.mdx

@@ -2,10 +2,11 @@
 title: Jina AI
 ---
 
-To use Jina AI embeddings, you need to import `JinaAIEmbedding` from `llamaindex`.
+To use Jina AI embeddings, you need to import `JinaAIEmbedding` from `@llamaindex/jinaai`.
 
 ```ts
-import { JinaAIEmbedding, Settings } from "llamaindex";
+import { Settings } from "llamaindex";
+import { JinaAIEmbedding } from "@llamaindex/jinaai";
 
 Settings.embedModel = new JinaAIEmbedding();
 

apps/next/src/content/docs/llamaindex/modules/embeddings/available_embeddings/together.mdx

@@ -2,10 +2,11 @@
 title: Together
 ---
 
-To use together embeddings, you need to import `TogetherEmbedding` from `llamaindex`.
+To use together embeddings, you need to import `TogetherEmbedding` from `@llamaindex/together`.
 
 ```ts
-import { TogetherEmbedding, Settings } from "llamaindex";
+import { Settings } from "llamaindex";
+import { TogetherEmbedding } from "@llamaindex/together";
 
 Settings.embedModel = new TogetherEmbedding({
   apiKey: "<YOUR_API_KEY>",

apps/next/src/content/docs/llamaindex/modules/llms/available_llms/deepseek.mdx

@@ -7,7 +7,8 @@ title: DeepSeek LLM
 ## Usage
 
 ```ts
-import { DeepSeekLLM, Settings } from "llamaindex";
+import { Settings } from "llamaindex";
+import { DeepSeekLLM } from "@llamaindex/deepseek";
 
 Settings.llm = new DeepSeekLLM({
   apiKey: "<YOUR_API_KEY>",
@@ -18,7 +19,8 @@ Settings.llm = new DeepSeekLLM({
 ## Example
 
 ```ts
-import { DeepSeekLLM, Document, VectorStoreIndex, Settings } from "llamaindex";
+import { Document, VectorStoreIndex, Settings } from "llamaindex";
+import { DeepSeekLLM } from "@llamaindex/deepseek";
 
 const deepseekLlm = new DeepSeekLLM({
   apiKey: "<YOUR_API_KEY>",

apps/next/src/content/docs/llamaindex/modules/llms/available_llms/fireworks.mdx

@@ -7,7 +7,8 @@ title: Fireworks LLM
 ## Usage
 
 ```ts
-import { FireworksLLM, Settings } from "llamaindex";
+import { Settings } from "llamaindex";
+import { FireworksLLM } from "@llamaindex/fireworks";
 
 Settings.llm = new FireworksLLM({
   apiKey: "<YOUR_API_KEY>",

apps/next/src/content/docs/llamaindex/modules/llms/available_llms/together.mdx

@@ -23,7 +23,8 @@ import { Tab, Tabs } from "fumadocs-ui/components/tabs";
 ## Usage
 
 ```ts
-import { Settings, TogetherLLM } from "llamaindex";
+import { Settings } from "llamaindex";
+import { TogetherLLM } from "@llamaindex/together";
 
 Settings.llm = new TogetherLLM({
   apiKey: "<YOUR_API_KEY>",

examples/deepseek.ts

+import { DeepSeekLLM } from "@llamaindex/deepseek";
+
+// process.env.DEEPSEEK_API_KEY is required
+const deepseek = new DeepSeekLLM({
+  apiKey: process.env.DEEPSEEK_API_KEY,
+  model: "deepseek-coder", // or "deepseek-chat"
+});
+
+(async () => {
+  // Example of non-streaming chat
+  const response = await deepseek.chat({
+    messages: [
+      {
+        role: "system",
+        content: "You are an AI assistant",
+      },
+      {
+        role: "user",
+        content: "Tell me about San Francisco",
+      },
+    ],
+    stream: false,
+  });
+  console.log("Response from DeepSeek AI:");
+  console.log(response);
+
+  // Example of streaming chat
+  const generator = await deepseek.chat({
+    messages: [
+      {
+        role: "system",
+        content: "You are an AI assistant",
+      },
+      {
+        role: "user",
+        content: "Write a short poem about San Francisco",
+      },
+    ],
+    stream: true,
+  });
+  console.log("\nStreaming response from DeepSeek AI...");
+  for await (const message of generator) {
+    process.stdout.write(message.delta);
+  }
+  console.log("\n");
+})();

examples/multimodal/jina.ts

+import { JinaAIEmbedding } from "@llamaindex/jinaai";
 import { SimpleDirectoryReader } from "@llamaindex/readers/directory";
-import {
-  ImageDocument,
-  JinaAIEmbedding,
-  similarity,
-  SimilarityType,
-} from "llamaindex";
+import { ImageDocument, similarity, SimilarityType } from "llamaindex";
 import path from "path";
 
 async function main() {

examples/package.json

@@ -44,6 +44,10 @@
     "@llamaindex/voyage-ai": "^1.0.3",
     "@llamaindex/weaviate": "^0.0.11",
     "@llamaindex/workflow": "^0.0.13",
+    "@llamaindex/deepseek": "^0.0.1",
+    "@llamaindex/fireworks": "^0.0.1",
+    "@llamaindex/together": "^0.0.1",
+    "@llamaindex/jinaai": "^0.0.1",
     "@notionhq/client": "^2.2.15",
     "@pinecone-database/pinecone": "^4.0.0",
     "@vercel/postgres": "^0.10.0",

examples/qdrantdb/with-gemini.ts

+import {
+  GEMINI_EMBEDDING_MODEL,
+  GeminiEmbedding,
+  GeminiSession,
+} from "@llamaindex/google";
+import { QdrantVectorStore } from "@llamaindex/qdrant";
+import {
+  Document,
+  storageContextFromDefaults,
+  VectorStoreIndex,
+} from "llamaindex";
+
+const embedding = new GeminiEmbedding({
+  model: GEMINI_EMBEDDING_MODEL.EMBEDDING_001,
+  session: new GeminiSession({
+    apiKey: process.env.GEMINI_API_KEY,
+  }),
+});
+
+async function main() {
+  const docs = [new Document({ text: "Lorem ipsum dolor sit amet" })];
+  const vectorStore = new QdrantVectorStore({
+    url: process.env.QDRANT_URL,
+    apiKey: process.env.QDRANT_API_KEY,
+    embeddingModel: embedding,
+    collectionName: "gemini_test",
+  });
+  const storageContext = await storageContextFromDefaults({ vectorStore });
+  await VectorStoreIndex.fromDocuments(docs, { storageContext });
+  console.log("Inizialized vector store successfully");
+}
+
+void main().catch((err) => console.error(err));

examples/qdrantdb/with-jina.ts

+import { JinaAIEmbedding } from "@llamaindex/jinaai";
+import { QdrantVectorStore } from "@llamaindex/qdrant";
+import {
+  Document,
+  storageContextFromDefaults,
+  VectorStoreIndex,
+} from "llamaindex";
+
+const embedding = new JinaAIEmbedding({
+  apiKey: process.env.JINAAI_API_KEY,
+  model: "jina-embeddings-v3",
+});
+
+async function main() {
+  const docs = [new Document({ text: "Lorem ipsum dolor sit amet" })];
+  const vectorStore = new QdrantVectorStore({
+    url: process.env.QDRANT_URL,
+    apiKey: process.env.QDRANT_API_KEY,
+    embeddingModel: embedding,
+    collectionName: "jina_test",
+  });
+  const storageContext = await storageContextFromDefaults({ vectorStore });
+  await VectorStoreIndex.fromDocuments(docs, { storageContext });
+  console.log("Inizialized vector store successfully");
+}
+
+void main().catch((err) => console.error(err));

examples/readers/src/pdf_fw.ts

+import { FireworksEmbedding, FireworksLLM } from "@llamaindex/fireworks";
 import { PDFReader } from "@llamaindex/readers/pdf";
-import { FireworksEmbedding, FireworksLLM, VectorStoreIndex } from "llamaindex";
+import { VectorStoreIndex } from "llamaindex";
 
 import { Settings } from "llamaindex";
 

examples/together-ai.ts

-import { TogetherEmbedding, TogetherLLM } from "llamaindex";
+import { TogetherEmbedding, TogetherLLM } from "@llamaindex/together";
 
 // process.env.TOGETHER_API_KEY is required
 const together = new TogetherLLM({

examples/together-ai/vector-index.ts

 import fs from "node:fs/promises";
 
-import {
-  Document,
-  Settings,
-  TogetherEmbedding,
-  TogetherLLM,
-  VectorStoreIndex,
-} from "llamaindex";
+import { TogetherEmbedding, TogetherLLM } from "@llamaindex/together";
+import { Document, Settings, VectorStoreIndex } from "llamaindex";
 
 // Update llm to use TogetherAI
 Settings.llm = new TogetherLLM({

packages/llamaindex/src/embeddings/OpenAIEmbedding.ts

-export * from "@llamaindex/openai";

packages/llamaindex/src/embeddings/index.ts

-export * from "@llamaindex/core/embeddings";
-export { FireworksEmbedding } from "./fireworks.js";
-export * from "./JinaAIEmbedding.js";
-export * from "./OpenAIEmbedding.js";
-export { TogetherEmbedding } from "./together.js";

packages/llamaindex/src/index.edge.ts

@@ -22,6 +22,7 @@ export {
 export * from "@llamaindex/core/agent";
 export * from "@llamaindex/core/chat-engine";
 export * from "@llamaindex/core/data-structs";
+export * from "@llamaindex/core/embeddings";
 export {
   CallbackManager,
   DEFAULT_BASE_URL,
@@ -65,10 +66,10 @@ export * from "@llamaindex/core/storage/doc-store";
 export * from "@llamaindex/core/storage/index-store";
 export * from "@llamaindex/core/storage/kv-store";
 export * from "@llamaindex/core/utils";
+export * from "@llamaindex/openai";
 export * from "@llamaindex/workflow/agent";
 export * from "./agent/index.js";
 export * from "./cloud/index.js";
-export * from "./embeddings/index.js";
 export * from "./engines/chat/index.js";
 export * from "./engines/query/index.js";
 export * from "./evaluation/index.js";
@@ -76,7 +77,6 @@ export * from "./extractors/index.js";
 export * from "./indices/index.js";
 export * from "./ingestion/index.js";
 export { imageToDataUrl } from "./internal/utils.js";
-export * from "./llm/index.js";
 export * from "./node-parser.js";
 export * from "./objects/index.js";
 export * from "./OutputParser.js";

packages/llamaindex/src/index.ts

 export * from "./index.edge.js";
 
-// TODO: clean up, move to jinaai package
-export { JinaAIEmbedding } from "./embeddings/JinaAIEmbedding.js";
-
 // Don't export file-system stores for non-node.js runtime on top level,
 //  as we cannot guarantee that they will work in other environments
 export * from "./storage/index.js";

packages/llamaindex/src/indices/vectorStore/index.ts

@@ -175,7 +175,7 @@ export class VectorStoreIndex extends BaseIndex<IndexDict> {
     for (const type in nodeMap) {
       const nodes = nodeMap[type as ModalityType];
       const embedModel =
-        this.embedModel ?? this.vectorStores[type as ModalityType]?.embedModel;
+        this.vectorStores[type as ModalityType]?.embedModel ?? this.embedModel;
       if (embedModel && nodes) {
         await embedModel(nodes, {
           logProgress: options?.logProgress,

packages/llamaindex/src/llm/index.ts

-export { DeepSeekLLM } from "./deepseek.js";
-export { FireworksLLM } from "./fireworks.js";
-export * from "./openai.js";
-export { TogetherLLM } from "./together.js";