fix:

feat: fix type
feat: workflow-based llama-reader
2026-07-19 18:43:34 -04:00 · 2025-04-08 13:34:14 -07:00 · 2025-04-08 13:22:36 -07:00 · 2025-04-08 12:28:30 -07:00 · 2025-04-07 00:50:16 -07:00 · 2025-04-05 18:41:39 +07:00
159 changed files with 4115 additions and 822 deletions
@@ -0,0 +1,6 @@
+---
+"llamaindex": patch
+"@llamaindex/workflow": patch
+---
+
+feat: support multi-resolution compatibility
@@ -1,5 +1,28 @@
 # @llamaindex/doc

+## 0.2.5
+
+### Patch Changes
+
+- 4999df1: bump nextjs
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.2.4
+
+### Patch Changes
+
+- 9c63f3f: Add support for openai responses api
+- Updated dependencies [9c63f3f]
+- Updated dependencies [c515a32]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+  - @llamaindex/workflow@1.0.2
+  - llamaindex@0.9.15
+  - @llamaindex/cloud@4.0.2
+  - @llamaindex/node-parser@2.0.2
+  - @llamaindex/readers@3.0.2
+
 ## 0.2.3

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/doc",
-  "version": "0.2.3",
+  "version": "0.2.5",
  "private": true,
  "scripts": {
    "postinstall": "fumadocs-mdx",
@@ -46,7 +46,7 @@
    "hast-util-to-jsx-runtime": "^2.3.2",
    "llamaindex": "workspace:*",
    "lucide-react": "^0.460.0",
-    "next": "^15.2.1",
+    "next": "^15.2.4",
    "next-themes": "^0.4.3",
    "react": "^19.0.0",
    "react-dom": "^19.0.0",
@@ -58,7 +58,7 @@ FunctionTool.from<Input>((input) => {
 {
  compilerOptions: {
    // ⬇️ add this line to your tsconfig.json
-    moduleResolution: "bundler", // or "node16"
+    moduleResolution: "bundler", // or "nodenext" | "node16" | "node"
  },
 }
 ```
@@ -89,6 +89,22 @@ In old module resolution, TypeScript will not be able to find the module because
 See more about [moduleResolution](https://www.typescriptlang.org/docs/handbook/modules/theory.html#module-resolution) or
 [TypeScript 5.0 blog](https://devblogs.microsoft.com/typescript/announcing-typescript-5-0/#--moduleresolution-bundler7).

+Note for `"moduleResolution": "node"`:
+
+`moduleResolution: node` does not strictly follow package.json exports. We still support `moduleResolution: node` with a limited import path. When using `moduleResolution: node`, you may encounter type errors when importing like this:
+
+```ts
+import { Settings } from "llamaindex/Settings";
+
+// Cannot find module 'llamaindex/Settings' or its corresponding type declarations.
+// There are types at '.../node_modules/llamaindex/dist/type/Settings.d.ts', but this result could not be resolved under your current 'moduleResolution' setting. Consider updating to 'node16', 'nodenext', or 'bundler'.
+```
+
+The simplest way to fix this without changing `moduleResolution` is to import directly from `llamaindex`:
+
+```ts
+import { Settings } from "llamaindex";
+```

 	</Accordion>
 </Accordions>
@@ -46,6 +46,156 @@ or
 Settings.llm = new OpenAI({ model: "gpt-3.5-turbo", temperature: 0, apiKey: <YOUR_API_KEY>, baseURL: "https://api.scaleway.ai/v1" });
 ```

+## Using OpenAI Responses API
+
+The OpenAI Responses API provides enhanced functionality for handling complex interactions, including built-in tools, annotations, and streaming responses. Here's how to use it:
+
+### Basic Setup
+
+```ts
+import { openaiResponses } from "@llamaindex/openai";
+
+const llm = openaiResponses({
+  model: "gpt-4o",
+  temperature: 0.1,
+  maxOutputTokens: 1000
+});
+```
+
+### Message Content Types
+
+The API supports different types of message content, including text and images:
+
+```ts
+const response = await llm.chat({
+  messages: [
+    {
+      role: "user",
+      content: [
+        {
+          type: "input_text",
+          text: "What's in this image?"
+        },
+        {
+          type: "input_image",
+          image_url: "https://example.com/image.jpg",
+          detail: "auto" // Optional: can be "auto", "low", or "high"
+        }
+      ]
+    }
+  ]
+});
+```
+
+### Advanced Features
+
+#### Built-in Tools
+
+```ts
+const llm = openaiResponses({
+  model: "gpt-4o",
+  builtInTools: [
+    {
+      type: "function",
+      name: "search_files",
+      description: "Search through available files"
+    }
+  ],
+  strict: true // Enable strict mode for tool calls
+});
+```
+
+#### Response Tracking and Storage
+
+```ts
+const llm = openaiResponses({
+  trackPreviousResponses: true, // Enable response tracking
+  store: true, // Store responses for future reference
+  user: "user-123", // Associate responses with a user
+  callMetadata: { // Add custom metadata
+    sessionId: "session-123",
+    context: "customer-support"
+  }
+});
+```
+
+#### Streaming Responses
+
+```ts
+const response = await llm.chat({
+  messages: [
+    {
+      role: "user",
+      content: "Generate a long response"
+    }
+  ],
+  stream: true // Enable streaming
+});
+
+for await (const chunk of response) {
+  console.log(chunk.delta); // Process each chunk of the response
+}
+```
+
+### Configuration Options
+
+The OpenAI Responses API supports various configuration options:
+
+```ts
+const llm = openaiResponses({
+  // Model and basic settings
+  model: "gpt-4o",
+  temperature: 0.1,
+  topP: 1,
+  maxOutputTokens: 1000,
+  
+  // API configuration
+  apiKey: "your-api-key",
+  baseURL: "custom-endpoint",
+  maxRetries: 10,
+  timeout: 60000,
+  
+  // Response handling
+  trackPreviousResponses: false,
+  store: false,
+  strict: false,
+  
+  // Additional options
+  instructions: "Custom instructions for the model",
+  truncation: "auto", // Can be "auto", "disabled", or null
+  include: ["citations", "reasoning"] // Specify what to include in responses
+});
+```
+
+### Response Structure
+
+The API returns responses with rich metadata and optional annotations:
+
+```ts
+interface ResponseStructure {
+  message: {
+    content: string;
+    role: "assistant";
+    options: {
+      built_in_tool_calls: Array<ToolCall>;
+      annotations?: Array<Citation | URLCitation | FilePath>;
+      refusal?: string;
+      reasoning?: ReasoningItem;
+      usage?: ResponseUsage;
+      toolCall?: Array<PartialToolCall>;
+    }
+  }
+}
+```
+
+### Best Practices
+
+1. Use `trackPreviousResponses` when you need conversation continuity
+2. Enable `strict` mode when using tools to ensure accurate function calls
+3. Set appropriate `maxOutputTokens` to control response length
+4. Use `annotations` to track citations and references in responses
+5. Implement error handling for potential API failures and retries
+
 ## Using JSON Response Format

 You can configure OpenAI to return responses in JSON format:
@@ -73,6 +223,112 @@ Settings.llm = new OpenAI({
 });
 ```

+## Response Formats
+
+The OpenAI LLM supports different response formats to structure the output in specific ways. There are two main approaches to formatting responses:
+
+### 1. JSON Object Format
+
+The simplest way to get structured JSON responses is using the `json_object` response format:
+
+```ts
+Settings.llm = new OpenAI({ 
+  model: "gpt-4o", 
+  temperature: 0,
+  responseFormat: { type: "json_object" }  
+});
+
+const response = await llm.chat({
+  messages: [
+    {
+      role: "system",
+      content: "You are a helpful assistant that outputs JSON."
+    },
+    {
+      role: "user", 
+      content: "Summarize this meeting transcript"
+    }
+  ]
+});
+
+// Response will be valid JSON
+console.log(response.message.content);
+```
+
+### 2. Schema Validation with Zod
+
+For more robust type safety and validation, you can use Zod schemas to define the expected response structure:
+
+```ts
+import { z } from "zod";
+
+// Define the response schema
+const meetingSchema = z.object({
+  summary: z.string(),
+  participants: z.array(z.string()),
+  actionItems: z.array(z.string()),
+  nextSteps: z.string()
+});
+
+// Configure the LLM with the schema
+Settings.llm = new OpenAI({ 
+  model: "gpt-4o", 
+  temperature: 0,
+  responseFormat: meetingSchema
+});
+
+const response = await llm.chat({
+  messages: [
+    {
+      role: "user",
+      content: "Summarize this meeting transcript" 
+    }
+  ]
+});
+
+// Response will be typed and validated according to the schema
+const result = response.message.content;
+console.log(result.summary);
+console.log(result.actionItems);
+```
+
+### Response Format Options
+
+The response format can be configured in two ways:
+
+1. At LLM initialization:
+```ts
+const llm = new OpenAI({
+  model: "gpt-4o",
+  responseFormat: { type: "json_object" } // or a Zod schema
+});
+```
+
+2. Per request:
+```ts
+const response = await llm.chat({
+  messages: [...],
+  responseFormat: { type: "json_object" } // or a Zod schema
+});
+```
+
+The response format options are:
+
+- `{ type: "json_object" }` - Returns responses as JSON objects
+- `zodSchema` - A Zod schema that defines and validates the response structure
+
+### Best Practices
+
+1. Use JSON object format for simple structured responses
+2. Use Zod schemas when you need:
+   - Type safety
+   - Response validation
+   - Complex nested structures
+   - Specific field constraints
+3. Set a low temperature (e.g. 0) when using structured outputs for more reliable formatting
+4. Include clear instructions in system or user messages about the expected response format
+5. Handle potential parsing errors when working with JSON responses
+
 ## Load and index documents

 For this example, we will use a single document. In a real-world scenario, you would have multiple documents to index.
@@ -1,5 +1,18 @@
 # @llamaindex/cloudflare-worker-agent-test

+## 0.0.150
+
+### Patch Changes
+
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.0.149
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.0.148

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.148",
+  "version": "0.0.150",
  "type": "module",
  "private": true,
  "scripts": {
@@ -1,5 +1,11 @@
 # @llamaindex/llama-parse-browser-test

+## 0.0.57
+
+### Patch Changes
+
+- @llamaindex/cloud@4.0.2
+
 ## 0.0.56

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/llama-parse-browser-test",
  "private": true,
-  "version": "0.0.56",
+  "version": "0.0.57",
  "type": "module",
  "scripts": {
    "dev": "vite",
@@ -1,5 +1,19 @@
 # @llamaindex/next-agent-test

+## 0.1.150
+
+### Patch Changes
+
+- 4999df1: bump nextjs
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.1.149
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.1.148

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-agent-test",
-  "version": "0.1.148",
+  "version": "0.1.150",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -10,7 +10,7 @@
  "dependencies": {
    "ai": "^4.0.0",
    "llamaindex": "workspace:*",
-    "next": "15.2.0",
+    "next": "^15.2.4",
    "react": "19.0.0",
    "react-dom": "19.0.0"
  },
@@ -1,5 +1,19 @@
 # test-edge-runtime

+## 0.1.149
+
+### Patch Changes
+
+- 4999df1: bump nextjs
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.1.148
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.1.147

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.147",
+  "version": "0.1.149",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -9,7 +9,7 @@
  },
  "dependencies": {
    "llamaindex": "workspace:*",
-    "next": "15.2.0",
+    "next": "^15.2.4",
    "react": "^19.0.0",
    "react-dom": "^19.0.0"
  },
@@ -1,5 +1,21 @@
 # @llamaindex/next-node-runtime

+## 0.1.16
+
+### Patch Changes
+
+- 4999df1: bump nextjs
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.1.15
+
+### Patch Changes
+
+- llamaindex@0.9.15
+- @llamaindex/huggingface@0.1.2
+- @llamaindex/readers@3.0.2
+
 ## 0.1.14

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.1.14",
+  "version": "0.1.16",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -11,7 +11,7 @@
    "@llamaindex/huggingface": "workspace:*",
    "@llamaindex/readers": "workspace:*",
    "llamaindex": "workspace:*",
-    "next": "15.2.0",
+    "next": "^15.2.4",
    "react": "19.0.0",
    "react-dom": "19.0.0"
  },
@@ -1,5 +1,18 @@
 # vite-import-llamaindex

+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.0.15
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.0.14

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "vite-import-llamaindex",
  "private": true,
-  "version": "0.0.14",
+  "version": "0.0.16",
  "type": "module",
  "scripts": {
    "build": "vite build",
@@ -1,5 +1,18 @@
 # @llamaindex/waku-query-engine-test

+## 0.0.150
+
+### Patch Changes
+
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.0.149
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.0.148

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.148",
+  "version": "0.0.150",
  "type": "module",
  "private": true,
  "scripts": {
@@ -1,5 +1,63 @@
 # examples

+## 0.3.3
+
+### Patch Changes
+
+- Updated dependencies [f5e4d09]
+- Updated dependencies [ead657a]
+  - llamaindex@0.9.16
+  - @llamaindex/tools@0.0.5
+
+## 0.3.2
+
+### Patch Changes
+
+- 9c63f3f: Add support for openai responses api
+- Updated dependencies [9c63f3f]
+- Updated dependencies [c515a32]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/google@0.2.2
+  - @llamaindex/core@0.6.2
+  - @llamaindex/workflow@1.0.2
+  - llamaindex@0.9.15
+  - @llamaindex/clip@0.0.48
+  - @llamaindex/deepinfra@0.0.48
+  - @llamaindex/deepseek@0.0.8
+  - @llamaindex/fireworks@0.0.8
+  - @llamaindex/groq@0.0.63
+  - @llamaindex/huggingface@0.1.2
+  - @llamaindex/jinaai@0.0.8
+  - @llamaindex/perplexity@0.0.5
+  - @llamaindex/azure@0.1.11
+  - @llamaindex/elastic-search@0.1.2
+  - @llamaindex/milvus@0.1.11
+  - @llamaindex/qdrant@0.1.11
+  - @llamaindex/supabase@0.1.1
+  - @llamaindex/together@0.0.8
+  - @llamaindex/vllm@0.0.34
+  - @llamaindex/cloud@4.0.2
+  - @llamaindex/node-parser@2.0.2
+  - @llamaindex/anthropic@0.3.2
+  - @llamaindex/cohere@0.0.16
+  - @llamaindex/mistral@0.1.2
+  - @llamaindex/mixedbread@0.0.16
+  - @llamaindex/ollama@0.1.2
+  - @llamaindex/portkey-ai@0.0.44
+  - @llamaindex/replicate@0.0.44
+  - @llamaindex/astra@0.0.16
+  - @llamaindex/chroma@0.0.16
+  - @llamaindex/firestore@1.0.9
+  - @llamaindex/mongodb@0.0.16
+  - @llamaindex/pinecone@0.1.2
+  - @llamaindex/postgres@0.0.44
+  - @llamaindex/upstash@0.0.16
+  - @llamaindex/weaviate@0.0.16
+  - @llamaindex/vercel@0.1.2
+  - @llamaindex/voyage-ai@1.0.8
+  - @llamaindex/readers@3.0.2
+  - @llamaindex/tools@0.0.4
+
 ## 0.3.1

 ### Patch Changes
@@ -0,0 +1,34 @@
+import { openai } from "@llamaindex/openai";
+import { mcp } from "@llamaindex/tools";
+import { agent } from "llamaindex";
+
+async function main() {
+  // Create an MCP server for filesystem tools
+  const server = mcp({
+    command: "npx",
+    args: ["-y", "@modelcontextprotocol/server-filesystem", "."],
+    verbose: true,
+  });
+
+  try {
+    // Create an agent that uses the MCP tools
+    const myAgent = agent({
+      name: "Assistant",
+      systemPrompt: "Use the tools to achieve the task.",
+      tools: await server.tools(),
+      llm: openai({ model: "gpt-4o" }),
+      verbose: true,
+    });
+
+    // Run a task
+    const response = await myAgent.run(
+      "what are the files in the current directory?",
+    );
+
+    console.log("Agent response:", response.data);
+  } finally {
+    await server.cleanup();
+  }
+}
+
+main().catch(console.error);
@@ -0,0 +1,30 @@
+import { openaiResponses } from "@llamaindex/openai";
+import { wiki } from "@llamaindex/tools";
+import { agent, tool } from "llamaindex";
+import { z } from "zod";
+
+const workflow = agent({
+  tools: [
+    tool({
+      name: "weather",
+      description: "Get the weather",
+      parameters: z.object({
+        location: z.string().describe("The location to get the weather for"),
+      }),
+      execute: ({ location }) => `The weather in ${location} is sunny`,
+    }),
+    wiki(),
+  ],
+  llm: openaiResponses({
+    model: "gpt-4o-mini",
+  }),
+});
+
+async function main() {
+  const result = await workflow.run(
+    "What is the weather in New York? What's the history of New York from Wikipedia in 3 sentences?",
+  );
+  console.log(result.data);
+}
+
+void main();
@@ -0,0 +1,33 @@
+import { openaiResponses } from "@llamaindex/openai";
+
+async function main() {
+  const llm = openaiResponses({
+    model: "gpt-4o",
+    maxOutputTokens: 1000,
+    apiKey: process.env.MY_OPENAI_API_KEY,
+  });
+
+  const response = await llm.chat({
+    messages: [
+      {
+        role: "user",
+        content: [
+          {
+            type: "text",
+            text: "What's in this image? Describe it in detail.",
+          },
+          {
+            type: "image_url",
+            image_url: {
+              url: "https://storage.googleapis.com/cloud-samples-data/vision/face/faces.jpeg",
+            },
+          },
+        ],
+      },
+    ],
+  });
+
+  console.log("Single Image Analysis:", response.message.content);
+}
+
+main().catch(console.error);
@@ -0,0 +1,22 @@
+import { openaiResponses } from "@llamaindex/openai";
+
+async function main() {
+  const llm = openaiResponses({
+    model: "gpt-4o-mini",
+    temperature: 0.1,
+  });
+
+  // Basic chat example
+  const response = await llm.chat({
+    messages: [
+      {
+        role: "user",
+        content: "What is the capital of France?",
+      },
+    ],
+  });
+
+  console.log(response.message.content);
+}
+
+main().catch(console.error);
@@ -0,0 +1,26 @@
+import { openaiResponses } from "@llamaindex/openai";
+
+async function main() {
+  const llm = openaiResponses({
+    model: "gpt-4o-mini",
+    temperature: 0.1,
+  });
+
+  const stream = await llm.chat({
+    messages: [
+      { content: "You want to talk in rhymes.", role: "system" },
+      {
+        content:
+          "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
+        role: "user",
+      },
+    ],
+    stream: true,
+  });
+
+  for await (const chunk of stream) {
+    process.stdout.write(chunk.delta);
+  }
+}
+
+main().catch(console.error);
@@ -0,0 +1,37 @@
+import { openaiResponses } from "@llamaindex/openai";
+import { tool } from "llamaindex";
+
+import { z } from "zod";
+async function main() {
+  const weatherTool = tool({
+    name: "weather",
+    description: "Get the weather",
+    parameters: z.object({
+      location: z.string({
+        description: "The location to get the weather for",
+      }),
+    }),
+    execute: ({ location }) => {
+      return `The weather in ${location} is sunny`;
+    },
+  });
+
+  const llm = openaiResponses({
+    model: "gpt-4o-mini",
+    temperature: 0.1,
+  });
+
+  const response = await llm.chat({
+    messages: [
+      {
+        role: "user",
+        content: "What is the weather in New York?",
+      },
+    ],
+    tools: [weatherTool],
+  });
+
+  console.log(response.message.options);
+}
+
+main().catch(console.error);
@@ -0,0 +1,23 @@
+import { openaiResponses } from "@llamaindex/openai";
+
+async function main() {
+  const llm = openaiResponses({
+    model: "gpt-4o",
+    temperature: 0.1,
+    builtInTools: [{ type: "web_search_preview" }],
+  });
+
+  // Streaming chat example
+  const response = await llm.chat({
+    messages: [
+      {
+        role: "user",
+        content: "What are the latest developments in AI?",
+      },
+    ],
+  });
+
+  console.log(response.message.content);
+}
+
+main().catch(console.error);
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/examples",
-  "version": "0.3.1",
+  "version": "0.3.3",
  "private": true,
  "scripts": {
    "lint": "eslint .",
@@ -11,47 +11,47 @@
    "@azure/cosmos": "^4.1.1",
    "@azure/identity": "^4.4.1",
    "@azure/search-documents": "^12.1.0",
-    "@llamaindex/anthropic": "^0.3.1",
-    "@llamaindex/astra": "^0.0.15",
-    "@llamaindex/azure": "^0.1.10",
-    "@llamaindex/chroma": "^0.0.15",
-    "@llamaindex/clip": "^0.0.47",
-    "@llamaindex/cloud": "^4.0.1",
-    "@llamaindex/cohere": "^0.0.15",
-    "@llamaindex/core": "^0.6.1",
-    "@llamaindex/deepinfra": "^0.0.47",
+    "@llamaindex/anthropic": "^0.3.2",
+    "@llamaindex/astra": "^0.0.16",
+    "@llamaindex/azure": "^0.1.11",
+    "@llamaindex/chroma": "^0.0.16",
+    "@llamaindex/clip": "^0.0.48",
+    "@llamaindex/cloud": "^4.0.2",
+    "@llamaindex/cohere": "^0.0.16",
+    "@llamaindex/core": "^0.6.2",
+    "@llamaindex/deepinfra": "^0.0.48",
    "@llamaindex/env": "^0.1.29",
-    "@llamaindex/firestore": "^1.0.8",
-    "@llamaindex/google": "^0.2.1",
-    "@llamaindex/groq": "^0.0.62",
-    "@llamaindex/huggingface": "^0.1.1",
-    "@llamaindex/milvus": "^0.1.10",
-    "@llamaindex/mistral": "^0.1.1",
-    "@llamaindex/mixedbread": "^0.0.15",
-    "@llamaindex/mongodb": "^0.0.15",
-    "@llamaindex/elastic-search": "^0.1.1",
-    "@llamaindex/node-parser": "^2.0.1",
-    "@llamaindex/ollama": "^0.1.1",
-    "@llamaindex/openai": "^0.2.1",
-    "@llamaindex/pinecone": "^0.1.1",
-    "@llamaindex/portkey-ai": "^0.0.43",
-    "@llamaindex/postgres": "^0.0.43",
-    "@llamaindex/qdrant": "^0.1.10",
-    "@llamaindex/readers": "^3.0.1",
-    "@llamaindex/replicate": "^0.0.43",
-    "@llamaindex/upstash": "^0.0.15",
-    "@llamaindex/vercel": "^0.1.1",
-    "@llamaindex/vllm": "^0.0.33",
-    "@llamaindex/voyage-ai": "^1.0.7",
-    "@llamaindex/weaviate": "^0.0.15",
-    "@llamaindex/workflow": "^1.0.1",
-    "@llamaindex/deepseek": "^0.0.7",
-    "@llamaindex/fireworks": "^0.0.7",
-    "@llamaindex/together": "^0.0.7",
-    "@llamaindex/jinaai": "^0.0.7",
-    "@llamaindex/perplexity": "^0.0.4",
-    "@llamaindex/supabase": "^0.1.0",
-    "@llamaindex/tools": "^0.0.3",
+    "@llamaindex/firestore": "^1.0.9",
+    "@llamaindex/google": "^0.2.2",
+    "@llamaindex/groq": "^0.0.63",
+    "@llamaindex/huggingface": "^0.1.2",
+    "@llamaindex/milvus": "^0.1.11",
+    "@llamaindex/mistral": "^0.1.2",
+    "@llamaindex/mixedbread": "^0.0.16",
+    "@llamaindex/mongodb": "^0.0.16",
+    "@llamaindex/elastic-search": "^0.1.2",
+    "@llamaindex/node-parser": "^2.0.2",
+    "@llamaindex/ollama": "^0.1.2",
+    "@llamaindex/openai": "^0.3.0",
+    "@llamaindex/pinecone": "^0.1.2",
+    "@llamaindex/portkey-ai": "^0.0.44",
+    "@llamaindex/postgres": "^0.0.44",
+    "@llamaindex/qdrant": "^0.1.11",
+    "@llamaindex/readers": "^3.0.2",
+    "@llamaindex/replicate": "^0.0.44",
+    "@llamaindex/upstash": "^0.0.16",
+    "@llamaindex/vercel": "^0.1.2",
+    "@llamaindex/vllm": "^0.0.34",
+    "@llamaindex/voyage-ai": "^1.0.8",
+    "@llamaindex/weaviate": "^0.0.16",
+    "@llamaindex/workflow": "^1.0.2",
+    "@llamaindex/deepseek": "^0.0.8",
+    "@llamaindex/fireworks": "^0.0.8",
+    "@llamaindex/together": "^0.0.8",
+    "@llamaindex/jinaai": "^0.0.8",
+    "@llamaindex/perplexity": "^0.0.5",
+    "@llamaindex/supabase": "^0.1.1",
+    "@llamaindex/tools": "^0.0.5",
    "@notionhq/client": "^2.2.15",
    "@pinecone-database/pinecone": "^4.0.0",
    "@vercel/postgres": "^0.10.0",
@@ -60,7 +60,7 @@
    "commander": "^12.1.0",
    "dotenv": "^16.4.5",
    "js-tiktoken": "^1.0.14",
-    "llamaindex": "^0.9.14",
+    "llamaindex": "^0.9.16",
    "mongodb": "6.7.0",
    "postgres": "^3.4.4",
    "wikipedia": "^2.1.2",
@@ -14,6 +14,7 @@
    "start:assemblyai": "node --import tsx ./src/assemblyai.ts",
    "start:llamaparse-dir": "node --import tsx ./src/simple-directory-reader-with-llamaparse.ts",
    "start:llamaparse-json": "node --import tsx ./src/llamaparse-json.ts",
+    "start:llamaparse-workflow": "node --import tsx ./src/llamaparse-workflow.ts",
    "start:discord": "node --import tsx ./src/discord.ts",
    "start:json": "node --import tsx ./src/json.ts",
    "start:obsidian": "node --import tsx ./src/obsidian.ts"
@@ -0,0 +1,26 @@
+import { uploadFile } from "@llamaindex/cloud/reader-workflow";
+
+async function main() {
+  const job = await uploadFile({
+    file: "../data/basic.pdf",
+  });
+  console.log("job id", job.jobId);
+
+  job.signal.addEventListener("abort", () => {
+    console.error("ERROR:", job.signal.reason);
+  });
+
+  const markdown = await job.markdown();
+  console.log("--markdown--");
+  console.log(markdown);
+
+  const text = await job.text();
+  console.log("--text--");
+  console.log(text);
+
+  const json = await job.json();
+  console.log("--json--");
+  console.log(json);
+}
+
+main();
@@ -35,7 +35,8 @@
    "prettier-plugin-tailwindcss": "^0.6.11",
    "turbo": "^2.4.4",
    "typescript": "^5.7.3",
-    "typescript-eslint": "^8.18.0"
+    "typescript-eslint": "^8.18.0",
+    "vitest": "^3.1.1"
  },
  "packageManager": "pnpm@9.12.3",
  "lint-staged": {
@@ -1,5 +1,19 @@
 # @llamaindex/autotool

+## 6.0.16
+
+### Patch Changes
+
+- 4999df1: bump nextjs
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 6.0.15
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 6.0.14

 ### Patch Changes
@@ -1,5 +1,21 @@
 # @llamaindex/autotool-01-node-example

+## 0.0.97
+
+### Patch Changes
+
+- Updated dependencies [4999df1]
+- Updated dependencies [f5e4d09]
+  - @llamaindex/autotool@6.0.16
+  - llamaindex@0.9.16
+
+## 0.0.96
+
+### Patch Changes
+
+- llamaindex@0.9.15
+- @llamaindex/autotool@6.0.15
+
 ## 0.0.95

 ### Patch Changes
@@ -13,5 +13,5 @@
  "scripts": {
    "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
  },
-  "version": "0.0.95"
+  "version": "0.0.97"
 }
@@ -6,7 +6,7 @@
    "url": "git+https://github.com/run-llama/LlamaIndexTS.git",
    "directory": "packages/autotool"
  },
-  "version": "6.0.14",
+  "version": "6.0.16",
  "description": "auto transpile your JS function to LLM Agent compatible",
  "files": [
    "dist",
@@ -77,7 +77,7 @@
    "@types/node": "^22.9.0",
    "bunchee": "6.4.0",
    "llamaindex": "workspace:*",
-    "next": "15.2.0",
+    "next": "^15.2.4",
    "rollup": "^4.28.1",
    "tsx": "^4.19.3",
    "typescript": "^5.7.3",
@@ -1,5 +1,12 @@
 # @llamaindex/cloud

+## 4.0.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 4.0.1

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloud",
-  "version": "4.0.1",
+  "version": "4.0.2",
  "type": "module",
  "license": "MIT",
  "scripts": {
@@ -47,6 +47,17 @@
        "default": "./reader/dist/index.js"
      },
      "default": "./reader/dist/index.js"
+    },
+    "./reader-workflow": {
+      "require": {
+        "types": "./reader/dist/workflow.d.cts",
+        "default": "./reader/dist/workflow.cjs"
+      },
+      "import": {
+        "types": "./reader/dist/workflow.d.ts",
+        "default": "./reader/dist/workflow.js"
+      },
+      "default": "./reader/dist/workflow.js"
    }
  },
  "repository": {
@@ -59,10 +70,16 @@
    "@hey-api/openapi-ts": "^0.61.0",
    "@llamaindex/core": "workspace:*",
    "@llamaindex/env": "workspace:*",
+    "@types/node": "^22.14.0",
    "bunchee": "6.4.0"
  },
  "peerDependencies": {
    "@llamaindex/core": "workspace:*",
    "@llamaindex/env": "workspace:*"
+  },
+  "dependencies": {
+    "fluere": "^0.3.3",
+    "stable-hash": "^0.0.5",
+    "zod": "^3.24.2"
  }
 }
@@ -0,0 +1,300 @@
+import { createClient, createConfig } from "@hey-api/client-fetch";
+import { fs, getEnv, path } from "@llamaindex/env";
+import {
+  createWorkflow,
+  type WorkflowEvent,
+  workflowEvent,
+  type WorkflowEventData,
+} from "fluere";
+import { withStore } from "fluere/middleware/store";
+import { withTraceEvents } from "fluere/middleware/trace-events";
+import { collect } from "fluere/stream/consumer";
+import { until } from "fluere/stream/until";
+import { pRetryHandler } from "fluere/util/p-retry";
+import { zodEvent } from "fluere/util/zod";
+import hash from "stable-hash";
+import { z } from "zod";
+import {
+  type Body_upload_file_api_v1_parsing_upload_post,
+  getJobApiV1ParsingJobJobIdGet,
+  getJobJsonResultApiV1ParsingJobJobIdResultJsonGet,
+  getJobResultApiV1ParsingJobJobIdResultMarkdownGet,
+  getJobTextResultApiV1ParsingJobJobIdResultTextGet,
+  type StatusEnum,
+  uploadFileApiV1ParsingUploadPost,
+} from "./api";
+import { parseFormSchema } from "./schema";
+
+type InferWorkflowEventData<T> =
+  T extends WorkflowEventData<infer U>
+    ? U
+    : T extends WorkflowEvent<infer U>
+      ? U
+      : never;
+
+const startEvent = zodEvent(
+  z.union([
+    parseFormSchema.merge(
+      z.object({
+        file: z
+          .string()
+          .or(z.instanceof(File))
+          .or(z.instanceof(Blob))
+          .or(z.instanceof(Uint8Array))
+          .optional()
+          .describe("input"),
+      }),
+    ),
+    parseFormSchema.merge(
+      z.object({
+        input_s3_path: z.string().optional(),
+      }),
+    ),
+    parseFormSchema.merge(
+      z.object({
+        input_url: z.string().optional(),
+      }),
+    ),
+  ]),
+);
+
+const checkStatusEvent = workflowEvent<string>();
+const checkStatusSuccessEvent = workflowEvent<string>();
+const requestMarkdownEvent = workflowEvent<string>();
+const requestTextEvent = workflowEvent<string>();
+const requestJsonEvent = workflowEvent<string>();
+
+const markdownResultEvent = workflowEvent<string>();
+const textResultEvent = workflowEvent<string>();
+const jsonResultEvent = workflowEvent<unknown>();
+
+export type LlamaParseWorkflowParams = {
+  region?: "us" | "eu" | "us-staging";
+  apiKey?: string;
+};
+
+const URLS = {
+  us: "https://api.cloud.llamaindex.ai",
+  eu: "https://api.cloud.eu.llamaindex.ai",
+  "us-staging": "https://api.staging.llamaindex.ai",
+} as const;
+
+const llamaParseWorkflow = withStore((params: LlamaParseWorkflowParams) => {
+  const apiKey = params.apiKey ?? getEnv("LLAMA_CLOUD_API_KEY");
+  const region = params.region ?? "us";
+  if (!apiKey) {
+    throw new Error("LLAMA_CLOUD_API_KEY is not set");
+  }
+  return {
+    cache: {} as Record<string, StatusEnum>,
+    client: createClient(
+      createConfig({
+        baseUrl: URLS[region],
+        headers: {
+          Authorization: `Bearer ${apiKey}`,
+        },
+      }),
+    ),
+  };
+}, withTraceEvents(createWorkflow()));
+
+llamaParseWorkflow.handle([startEvent], async ({ data: form }) => {
+  const store = llamaParseWorkflow.getStore();
+  const finalForm = { ...form } as Body_upload_file_api_v1_parsing_upload_post;
+  if ("file" in form) {
+    // support loads from the file system
+    const file = form?.file;
+    const isFilePath = typeof file === "string";
+    const data = isFilePath ? await fs.readFile(file) : file;
+    const filename: string | undefined = isFilePath
+      ? path.basename(file)
+      : undefined;
+    finalForm.file = data
+      ? globalThis.File && filename
+        ? new File([data], filename)
+        : new Blob([data])
+      : null;
+  }
+  const {
+    data: { id, status },
+  } = await uploadFileApiV1ParsingUploadPost({
+    throwOnError: true,
+    body: {
+      ...finalForm,
+    } satisfies {
+      [Key in keyof Body_upload_file_api_v1_parsing_upload_post]:
+        | Body_upload_file_api_v1_parsing_upload_post[Key]
+        | undefined;
+    } as Body_upload_file_api_v1_parsing_upload_post,
+    client: store.client,
+  });
+  store.cache[id] = status;
+  return checkStatusEvent.with(id);
+});
+
+llamaParseWorkflow.handle(
+  [checkStatusEvent],
+  pRetryHandler(
+    async ({ data: uuid }) => {
+      const store = llamaParseWorkflow.getStore();
+      if (store.cache[uuid] === "SUCCESS") {
+        {
+          return checkStatusSuccessEvent.with(uuid);
+        }
+      }
+      const {
+        data: { status },
+      } = await getJobApiV1ParsingJobJobIdGet({
+        throwOnError: true,
+        path: {
+          job_id: uuid,
+        },
+        client: store.client,
+      });
+      store.cache[uuid] = status;
+      if (status === "SUCCESS") {
+        return checkStatusSuccessEvent.with(uuid);
+      }
+      throw new Error(`LLamaParse status: ${status}`);
+    },
+    {
+      retries: 100,
+    },
+  ),
+);
+
+llamaParseWorkflow.handle([requestMarkdownEvent], async ({ data: job_id }) => {
+  const store = llamaParseWorkflow.getStore();
+  const { data } = await getJobResultApiV1ParsingJobJobIdResultMarkdownGet({
+    throwOnError: true,
+    path: {
+      job_id,
+    },
+    client: store.client,
+  });
+  return markdownResultEvent.with(data.markdown);
+});
+
+llamaParseWorkflow.handle([requestTextEvent], async ({ data: job_id }) => {
+  const store = llamaParseWorkflow.getStore();
+  const { data } = await getJobTextResultApiV1ParsingJobJobIdResultTextGet({
+    throwOnError: true,
+    path: {
+      job_id,
+    },
+    client: store.client,
+  });
+  return textResultEvent.with(data.text);
+});
+
+llamaParseWorkflow.handle([requestJsonEvent], async ({ data: job_id }) => {
+  const store = llamaParseWorkflow.getStore();
+  const { data } = await getJobJsonResultApiV1ParsingJobJobIdResultJsonGet({
+    throwOnError: true,
+    path: {
+      job_id,
+    },
+    client: store.client,
+  });
+  return jsonResultEvent.with(data.pages);
+});
+
+const cacheMap = new Map<
+  string,
+  ReturnType<typeof llamaParseWorkflow.createContext>
+>();
+
+export type ParseJob = {
+  get jobId(): string;
+  get signal(): AbortSignal;
+  get context(): ReturnType<typeof llamaParseWorkflow.createContext>;
+  get form(): InferWorkflowEventData<typeof startEvent>;
+
+  markdown(): Promise<string>;
+  text(): Promise<string>;
+  //eslint-disable-next-line @typescript-eslint/no-explicit-any
+  json(): Promise<any[]>;
+};
+
+export const upload = async (
+  params: InferWorkflowEventData<typeof startEvent> & LlamaParseWorkflowParams,
+): Promise<ParseJob> => {
+  //#region cache
+  const key = hash({ apiKey: params.apiKey, region: params.region });
+  if (!cacheMap.has(key)) {
+    const context = llamaParseWorkflow.createContext(params);
+    cacheMap.set(key, context);
+  }
+  //#endregion
+
+  //#region upload event
+  const context = cacheMap.get(key)!;
+  const { stream, sendEvent, createFilter } = context;
+  const ev = startEvent.with(params);
+  sendEvent(ev);
+  const uploadThread = await collect(
+    until(
+      stream,
+      createFilter(ev, (ev) => checkStatusSuccessEvent.include(ev)),
+    ),
+  );
+  //#region
+  const jobId: string = uploadThread.at(-1)!.data;
+  const job = {
+    get signal() {
+      // lazy load
+      return context.signal;
+    },
+    get jobId() {
+      return jobId;
+    },
+    get form() {
+      return ev.data;
+    },
+    get context() {
+      return context;
+    },
+    async markdown(): Promise<string> {
+      const requestEv = requestMarkdownEvent.with(jobId);
+      sendEvent(requestEv);
+      const markdownThread = await collect(
+        until(
+          stream,
+          createFilter(requestEv, (ev) => markdownResultEvent.include(ev)),
+        ),
+      );
+      return markdownThread.at(-1)!.data;
+    },
+    async text(): Promise<string> {
+      const requestEv = requestTextEvent.with(jobId);
+      sendEvent(requestEv);
+      const textThread = await collect(
+        until(
+          stream,
+          createFilter(requestEv, (ev) => textResultEvent.include(ev)),
+        ),
+      );
+      return textThread.at(-1)!.data;
+    },
+    //eslint-disable-next-line @typescript-eslint/no-explicit-any
+    async json(): Promise<any[]> {
+      const requestEv = requestJsonEvent.with(jobId);
+      sendEvent(requestEv);
+      const jsonThread = await collect(
+        until(
+          stream,
+          createFilter(requestEv, (ev) => jsonResultEvent.include(ev)),
+        ),
+      );
+      return jsonThread.at(-1)!.data;
+    },
+    async images(): Promise<void> {
+      const json = await job.json();
+      const images = json.flatMap(({ images }) => images);
+      images.map((image) => {
+        // todo
+      });
+    },
+  };
+  return job;
+};
@@ -0,0 +1,121 @@
+import { ParserLanguages, ParsingMode } from "./api";
+
+import { z } from "zod";
+
+type Language = (typeof ParserLanguages)[keyof typeof ParserLanguages];
+const VALUES: [Language, ...Language[]] = [
+  ParserLanguages.EN,
+  ...Object.values(ParserLanguages),
+];
+const languageSchema = z.enum(VALUES);
+
+const PARSE_PRESETS = [
+  "fast",
+  "balanced",
+  "premium",
+  "structured",
+  "auto",
+  "scientific",
+  "invoice",
+  "slides",
+  "_carlyle",
+] as const;
+
+export const parsePresetSchema = z.enum(PARSE_PRESETS);
+
+export const parseFormSchema = z.object({
+  input_s3_region: z.string().optional(),
+  adaptive_long_table: z.boolean().optional(),
+  annotate_links: z.boolean().optional(),
+  auto_mode: z.boolean().optional(),
+  auto_mode_trigger_on_image_in_page: z.boolean().optional(),
+  auto_mode_trigger_on_table_in_page: z.boolean().optional(),
+  auto_mode_trigger_on_text_in_page: z.string().optional(),
+  auto_mode_trigger_on_regexp_in_page: z.string().optional(),
+  azure_openai_api_version: z.string().optional(),
+  azure_openai_deployment_name: z.string().optional(),
+  azure_openai_endpoint: z.string().optional(),
+  azure_openai_key: z.string().optional(),
+  bbox_bottom: z.number().min(0).max(1).optional(),
+  bbox_left: z.number().min(0).max(1).optional(),
+  bbox_right: z.number().min(0).max(1).optional(),
+  bbox_top: z.number().min(0).max(1).optional(),
+  disable_ocr: z.boolean().optional(),
+  disable_reconstruction: z.boolean().optional(),
+  disable_image_extraction: z.boolean().optional(),
+  do_not_cache: z.coerce.boolean().optional(),
+  do_not_unroll_columns: z.coerce.boolean().optional(),
+  extract_charts: z.boolean().optional(),
+  guess_xlsx_sheet_name: z.boolean().optional(),
+  html_make_all_elements_visible: z.boolean().optional(),
+  html_remove_fixed_elements: z.boolean().optional(),
+  html_remove_navigation_elements: z.boolean().optional(),
+  http_proxy: z
+    .string()
+    .url(
+      'Set a valid URL for the HTTP proxy, e.g., "http://proxy.example.com:8080"',
+    )
+    .refine(
+      (url) => {
+        try {
+          const parsedUrl = new URL(url);
+          return (
+            parsedUrl.protocol === "http:" || parsedUrl.protocol === "https:"
+          );
+        } catch {
+          return false;
+        }
+      },
+      {
+        message: "Invalid HTTP proxy URL",
+      },
+    )
+    .optional(),
+  invalidate_cache: z.boolean().optional(),
+  language: z.array(languageSchema).optional(),
+  extract_layout: z.boolean().optional(),
+  max_pages: z.number().nullable().optional(),
+  output_pdf_of_document: z.boolean().optional(),
+  output_s3_path_prefix: z.string().optional(),
+  output_s3_region: z.string().optional(),
+  page_prefix: z.string().optional(),
+  page_separator: z.string().optional(),
+  page_suffix: z.string().optional(),
+  preserve_layout_alignment_across_pages: z.boolean().optional(),
+  skip_diagonal_text: z.boolean().optional(),
+  spreadsheet_extract_sub_tables: z.boolean().optional(),
+  structured_output: z.boolean().optional(),
+  structured_output_json_schema: z.string().optional(),
+  structured_output_json_schema_name: z.string().optional(),
+  take_screenshot: z.boolean().optional(),
+  target_pages: z.string().optional(),
+  vendor_multimodal_api_key: z.string().optional(),
+  vendor_multimodal_model_name: z.string().optional(),
+  model: z.string().optional(),
+  webhook_url: z.string().url().optional(),
+  parse_mode: z.nativeEnum(ParsingMode).nullable().optional(),
+  system_prompt: z.string().optional(),
+  system_prompt_append: z.string().optional(),
+  user_prompt: z.string().optional(),
+  job_timeout_in_seconds: z.number().optional(),
+  job_timeout_extra_time_per_page_in_seconds: z.number().optional(),
+  strict_mode_image_extraction: z.boolean().optional(),
+  strict_mode_image_ocr: z.boolean().optional(),
+  strict_mode_reconstruction: z.boolean().optional(),
+  strict_mode_buggy_font: z.boolean().optional(),
+  ignore_document_elements_for_layout_detection: z.boolean().optional(),
+  output_tables_as_HTML: z.boolean().optional(),
+  use_vendor_multimodal_model: z.boolean().optional(),
+  bounding_box: z.string().optional(),
+  gpt4o_mode: z.boolean().optional(),
+  gpt4o_api_key: z.string().optional(),
+  complemental_formatting_instruction: z.string().optional(),
+  content_guideline_instruction: z.string().optional(),
+  premium_mode: z.boolean().optional(),
+  is_formatting_instruction: z.boolean().optional(),
+  continuous_mode: z.boolean().optional(),
+  parsing_instruction: z.string().optional(),
+  fast_mode: z.boolean().optional(),
+  formatting_instruction: z.string().optional(),
+  preset: parsePresetSchema.optional(),
+});
@@ -8,7 +8,7 @@
    "moduleResolution": "Bundler",
    "skipLibCheck": true,
    "strict": true,
-    "types": []
+    "types": ["node"]
  },
  "include": ["./src"],
  "exclude": ["node_modules"],
@@ -1,5 +1,12 @@
 # @llamaindex/community

+## 0.0.94
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.93

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/community",
  "description": "Community package for LlamaIndexTS",
-  "version": "0.0.93",
+  "version": "0.0.94",
  "type": "module",
  "types": "dist/type/index.d.ts",
  "main": "dist/cjs/index.js",
@@ -1,5 +1,11 @@
 # @llamaindex/core

+## 0.6.2
+
+### Patch Changes
+
+- 9c63f3f: Add support for openai responses api
+
 ## 0.6.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/core",
  "type": "module",
-  "version": "0.6.1",
+  "version": "0.6.2",
  "description": "LlamaIndex Core Module",
  "exports": {
    "./agent": {
@@ -2,7 +2,6 @@ import type { Tokenizers } from "@llamaindex/env/tokenizers";
 import type { JSONSchemaType } from "ajv";
 import { z } from "zod";
 import type { JSONObject, JSONValue } from "../global";
-
 /**
 * @internal
 */
@@ -55,7 +54,12 @@ export interface LLM<
  ): Promise<CompletionResponse>;
 }

-export type MessageType = "user" | "assistant" | "system" | "memory";
+export type MessageType =
+  | "user"
+  | "assistant"
+  | "system"
+  | "memory"
+  | "developer";

 export type TextChatMessage<AdditionalMessageOptions extends object = object> =
  {
@@ -156,6 +160,7 @@ export type MessageContentTextDetail = {
 export type MessageContentImageDetail = {
  type: "image_url";
  image_url: { url: string };
+  detail?: "high" | "low" | "auto";
 };

 export type MessageContentDetail =
@@ -1,5 +1,18 @@
 # @llamaindex/experimental

+## 0.0.166
+
+### Patch Changes
+
+- Updated dependencies [f5e4d09]
+  - llamaindex@0.9.16
+
+## 0.0.165
+
+### Patch Changes
+
+- llamaindex@0.9.15
+
 ## 0.0.164

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/experimental",
  "description": "Experimental package for LlamaIndexTS",
-  "version": "0.0.164",
+  "version": "0.0.166",
  "type": "module",
  "types": "dist/type/index.d.ts",
  "main": "dist/cjs/index.js",
@@ -1,5 +1,23 @@
 # llamaindex

+## 0.9.16
+
+### Patch Changes
+
+- f5e4d09: Remove gpt-tokenizer from llamaindex
+
+## 0.9.15
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+- Updated dependencies [c515a32]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+  - @llamaindex/workflow@1.0.2
+  - @llamaindex/cloud@4.0.2
+  - @llamaindex/node-parser@2.0.2
+
 ## 0.9.14

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "llamaindex",
-  "version": "0.9.14",
+  "version": "0.9.16",
  "license": "MIT",
  "type": "module",
  "keywords": [
@@ -29,7 +29,6 @@
    "@types/lodash": "^4.17.7",
    "@types/node": "^22.9.0",
    "ajv": "^8.17.1",
-    "gpt-tokenizer": "^2.6.2",
    "lodash": "^4.17.21",
    "magic-bytes.js": "^1.10.0"
  },
@@ -68,7 +68,6 @@ export * from "@llamaindex/core/storage/kv-store";
 export * from "@llamaindex/core/utils";
 export * from "@llamaindex/openai";
 export * from "@llamaindex/workflow";
-export * from "@llamaindex/workflow/agent";
 export * from "./agent/index.js";
 export * from "./cloud/index.js";
 export * from "./engines/chat/index.js";
@@ -1,5 +1,12 @@
 # @llamaindex/node-parser

+## 2.0.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 2.0.1

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/node-parser",
-  "version": "2.0.1",
+  "version": "2.0.2",
  "description": "Node parser for LlamaIndex",
  "type": "module",
  "exports": {
@@ -1,5 +1,12 @@
 # @llamaindex/anthropic

+## 0.3.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.3.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/anthropic",
  "description": "Anthropic Adapter for LlamaIndex",
-  "version": "0.3.1",
+  "version": "0.3.2",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,13 @@
 # @llamaindex/clip

+## 0.0.48
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+
 ## 0.0.47

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/clip",
  "description": "Clip Embedding Adapter for LlamaIndex",
-  "version": "0.0.47",
+  "version": "0.0.48",
  "type": "module",
  "types": "dist/index.d.ts",
  "main": "dist/index.cjs",
@@ -1,5 +1,12 @@
 # @llamaindex/cohere

+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.15

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/cohere",
  "description": "Cohere Adapter for LlamaIndex",
-  "version": "0.0.15",
+  "version": "0.0.16",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,13 @@
 # @llamaindex/deepinfra

+## 0.0.48
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+
 ## 0.0.47

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/deepinfra",
  "description": "Deepinfra Adapter for LlamaIndex",
-  "version": "0.0.47",
+  "version": "0.0.48",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/deepseek

+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+
 ## 0.0.7

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/deepseek",
  "description": "DeepSeek Adapter for LlamaIndex",
-  "version": "0.0.7",
+  "version": "0.0.8",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/fireworks

+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+
 ## 0.0.7

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/fireworks",
  "description": "Fireworks Adapter for LlamaIndex",
-  "version": "0.0.7",
+  "version": "0.0.8",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,13 @@
 # @llamaindex/google

+## 0.2.2
+
+### Patch Changes
+
+- 9c63f3f: Add support for openai responses api
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.2.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/google",
  "description": "Google Adapter for LlamaIndex",
-  "version": "0.2.1",
+  "version": "0.2.2",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -181,7 +181,7 @@ export const mapBaseToolToGeminiFunctionDeclaration = (
 export class GeminiHelper {
  // Gemini only has user and model roles. Put the rest in user role.
  public static readonly ROLES_TO_GEMINI: Record<
-    MessageType,
+    Exclude<MessageType, "developer">,
    GeminiMessageRole
  > = {
    user: "user",
@@ -285,7 +285,9 @@ export class GeminiHelper {
    if (message.options && "toolResult" in message.options) {
      return "function";
    }
-    return GeminiHelper.ROLES_TO_GEMINI[message.role];
+    return GeminiHelper.ROLES_TO_GEMINI[
+      message.role as Exclude<MessageType, "developer">
+    ];
  }

  public static chatMessageToGemini(
@@ -1,5 +1,12 @@
 # @llamaindex/groq

+## 0.0.63
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+
 ## 0.0.62

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/groq",
  "description": "Groq Adapter for LlamaIndex",
-  "version": "0.0.62",
+  "version": "0.0.63",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,13 @@
 # @llamaindex/huggingface

+## 0.1.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+
 ## 0.1.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/huggingface",
  "description": "Huggingface Adapter for LlamaIndex",
-  "version": "0.1.1",
+  "version": "0.1.2",
  "type": "module",
  "types": "dist/index.d.ts",
  "main": "dist/index.cjs",
@@ -1,5 +1,13 @@
 # @llamaindex/jinaai

+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+
 ## 0.0.7

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/jinaai",
  "description": "JinaAI Adapter for LlamaIndex",
-  "version": "0.0.7",
+  "version": "0.0.8",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/mistral

+## 0.1.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.1.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/mistral",
  "description": "Mistral Adapter for LlamaIndex",
-  "version": "0.1.1",
+  "version": "0.1.2",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/mixedbread

+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.15

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/mixedbread",
  "description": "Mixedbread Adapter for LlamaIndex",
-  "version": "0.0.15",
+  "version": "0.0.16",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/ollama

+## 0.1.2
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.1.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/ollama",
  "description": "Ollama Adapter for LlamaIndex",
-  "version": "0.1.1",
+  "version": "0.1.2",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,16 @@
 # @llamaindex/openai

+## 0.3.0
+
+### Minor Changes
+
+- 9c63f3f: Add support for openai responses api
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.2.1

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/openai",
  "description": "OpenAI Adapter for LlamaIndex",
-  "version": "0.2.1",
+  "version": "0.3.0",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -27,7 +27,8 @@
  },
  "scripts": {
    "build": "bunchee",
-    "dev": "bunchee --watch"
+    "dev": "bunchee --watch",
+    "test": "vitest"
  },
  "devDependencies": {
    "bunchee": "6.4.0"
@@ -35,7 +36,7 @@
  "dependencies": {
    "@llamaindex/core": "workspace:*",
    "@llamaindex/env": "workspace:*",
-    "openai": "^4.86.0",
+    "openai": "^4.90.0",
    "zod": "^3.24.2"
  }
 }
@@ -5,7 +5,8 @@ import {
 } from "@llamaindex/core/agent";
 import { Settings } from "@llamaindex/core/global";
 import type { ToolCallLLMMessageOptions } from "@llamaindex/core/llms";
-import { OpenAI, type OpenAIAdditionalChatOptions } from "./llm";
+import { OpenAI } from "./llm";
+import type { OpenAIAdditionalChatOptions } from "./utils";

 // This is likely not necessary anymore but leaving it here just in case it's in use elsewhere

@@ -1,3 +1,4 @@
 export * from "./agent";
 export * from "./embedding";
 export * from "./llm";
+export * from "./responses";
@@ -4,7 +4,6 @@ import {
  type ChatMessage,
  type ChatResponse,
  type ChatResponseChunk,
-  type LLM,
  type LLMChatParamsNonStreaming,
  type LLMChatParamsStreaming,
  type LLMMetadata,
@@ -18,7 +17,6 @@ import { getEnv } from "@llamaindex/env";
 import { Tokenizers } from "@llamaindex/env/tokenizers";
 import type {
  AzureClientOptions,
-  AzureOpenAI as AzureOpenAILLM,
  ClientOptions as OpenAIClientOptions,
  OpenAI as OpenAILLM,
 } from "openai";
@@ -45,153 +43,15 @@ import {
  getAzureModel,
  shouldUseAzure,
 } from "./azure.js";
-
-export const GPT4_MODELS = {
-  "chatgpt-4o-latest": {
-    contextWindow: 128000,
-  },
-  "gpt-4.5-preview": { contextWindow: 128000 },
-  "gpt-4.5-preview-2025-02-27": { contextWindow: 128000 },
-  "gpt-4": { contextWindow: 8192 },
-  "gpt-4-32k": { contextWindow: 32768 },
-  "gpt-4-32k-0613": { contextWindow: 32768 },
-  "gpt-4-turbo": { contextWindow: 128000 },
-  "gpt-4-turbo-preview": { contextWindow: 128000 },
-  "gpt-4-1106-preview": { contextWindow: 128000 },
-  "gpt-4-0125-preview": { contextWindow: 128000 },
-  "gpt-4-vision-preview": { contextWindow: 128000 },
-  "gpt-4o": { contextWindow: 128000 },
-  "gpt-4o-2024-05-13": { contextWindow: 128000 },
-  "gpt-4o-mini": { contextWindow: 128000 },
-  "gpt-4o-mini-2024-07-18": { contextWindow: 128000 },
-  "gpt-4o-2024-08-06": { contextWindow: 128000 },
-  "gpt-4o-2024-09-14": { contextWindow: 128000 },
-  "gpt-4o-2024-10-14": { contextWindow: 128000 },
-  "gpt-4-0613": { contextWindow: 128000 },
-  "gpt-4-turbo-2024-04-09": { contextWindow: 128000 },
-  "gpt-4-0314": { contextWindow: 128000 },
-  "gpt-4-32k-0314": { contextWindow: 32768 },
-  "gpt-4o-realtime-preview": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-realtime-preview-2024-10-01": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-audio-preview": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-audio-preview-2024-10-01": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-2024-11-20": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-audio-preview-2024-12-17": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-mini-audio-preview": {
-    contextWindow: 128000,
-  },
-  "gpt-4o-mini-audio-preview-2024-12-17": {
-    contextWindow: 128000,
-  },
-};
-
-// NOTE we don't currently support gpt-3.5-turbo-instruct and don't plan to in the near future
-export const GPT35_MODELS = {
-  "gpt-3.5-turbo": { contextWindow: 16385 },
-  "gpt-3.5-turbo-0613": { contextWindow: 4096 },
-  "gpt-3.5-turbo-16k": { contextWindow: 16385 },
-  "gpt-3.5-turbo-16k-0613": { contextWindow: 16385 },
-  "gpt-3.5-turbo-1106": { contextWindow: 16385 },
-  "gpt-3.5-turbo-0125": { contextWindow: 16385 },
-  "gpt-3.5-turbo-0301": { contextWindow: 16385 },
-};
-
-export const O1_MODELS = {
-  "o1-preview": {
-    contextWindow: 128000,
-  },
-  "o1-preview-2024-09-12": {
-    contextWindow: 128000,
-  },
-  "o1-mini": {
-    contextWindow: 128000,
-  },
-  "o1-mini-2024-09-12": {
-    contextWindow: 128000,
-  },
-  o1: {
-    contextWindow: 128000,
-  },
-  "o1-2024-12-17": {
-    contextWindow: 128000,
-  },
-};
-
-export const O3_MODELS = {
-  "o3-mini": {
-    contextWindow: 200000,
-  },
-  "o3-mini-2025-01-31": {
-    contextWindow: 200000,
-  },
-};
-
-/**
- * We currently support GPT-3.5 and GPT-4 models
- */
-export const ALL_AVAILABLE_OPENAI_MODELS = {
-  ...GPT4_MODELS,
-  ...GPT35_MODELS,
-  ...O1_MODELS,
-  ...O3_MODELS,
-} satisfies Record<ChatModel, { contextWindow: number }>;
-
-function isFunctionCallingModel(llm: LLM): llm is OpenAI {
-  let model: string;
-  if (llm instanceof OpenAI) {
-    model = llm.model;
-  } else if ("model" in llm && typeof llm.model === "string") {
-    model = llm.model;
-  } else {
-    return false;
-  }
-  const isChatModel = Object.keys(ALL_AVAILABLE_OPENAI_MODELS).includes(model);
-  const isOld = model.includes("0314") || model.includes("0301");
-  const isO1 = model.startsWith("o1");
-  return isChatModel && !isOld && !isO1;
-}
-
-function isReasoningModel(model: ChatModel | string): boolean {
-  const isO1 = model.startsWith("o1");
-  const isO3 = model.startsWith("o3");
-  return isO1 || isO3;
-}
-
-function isTemperatureSupported(model: ChatModel | string): boolean {
-  return !model.startsWith("o3");
-}
-
-export type OpenAIAdditionalMetadata = object;
-
-export type OpenAIAdditionalChatOptions = Omit<
-  Partial<OpenAILLM.Chat.ChatCompletionCreateParams>,
-  | "max_tokens"
-  | "messages"
-  | "model"
-  | "temperature"
-  | "reasoning_effort"
-  | "top_p"
-  | "stream"
-  | "tools"
-  | "toolChoice"
->;
-
-type LLMInstance = Pick<
-  AzureOpenAILLM | OpenAILLM,
-  "chat" | "apiKey" | "baseURL"
->;
+import {
+  ALL_AVAILABLE_OPENAI_MODELS,
+  isFunctionCallingModel,
+  isReasoningModel,
+  isTemperatureSupported,
+  type LLMInstance,
+  type OpenAIAdditionalChatOptions,
+  type OpenAIAdditionalMetadata,
+} from "./utils.js";

 export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
  model:
@@ -230,6 +90,7 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
    },
  ) {
    super();
+
    this.model = init?.model ?? "gpt-4o";
    this.temperature = init?.temperature ?? 0.1;
    this.reasoningEffort = isReasoningModel(this.model)
@@ -0,0 +1,774 @@
+import {
+  ToolCallLLM,
+  type BaseTool,
+  type ChatMessage,
+  type ChatResponse,
+  type ChatResponseChunk,
+  type LLMChatParamsNonStreaming,
+  type LLMChatParamsStreaming,
+  type LLMMetadata,
+  type MessageContent,
+  type MessageType,
+  type PartialToolCall,
+  type ToolCallLLMMessageOptions,
+  type ToolCallOptions,
+  type ToolResultOptions,
+} from "@llamaindex/core/llms";
+import type { StoredValue } from "@llamaindex/core/schema";
+import { extractText } from "@llamaindex/core/utils";
+import { getEnv } from "@llamaindex/env";
+import {
+  OpenAI as OpenAILLM,
+  type AzureClientOptions,
+  type ClientOptions as OpenAIClientOptions,
+} from "openai";
+
+import { wrapEventCaller } from "@llamaindex/core/decorator";
+import { Tokenizers } from "@llamaindex/env/tokenizers";
+import {
+  AzureOpenAIWithUserAgent,
+  getAzureConfigFromEnv,
+  getAzureModel,
+  shouldUseAzure,
+} from "./azure";
+import {
+  ALL_AVAILABLE_OPENAI_MODELS,
+  isFunctionCallingModel,
+  isReasoningModel,
+  isTemperatureSupported,
+  type LLMInstance,
+  type OpenAIAdditionalMetadata,
+  type OpenAIResponsesChatOptions,
+  type OpenAIResponsesRole,
+  type ResponseMessageContent,
+  type ResponsesAdditionalOptions,
+  type StreamState,
+} from "./utils";
+
+export class OpenAIResponses extends ToolCallLLM<OpenAIResponsesChatOptions> {
+  model: string;
+  temperature: number;
+  topP: number;
+  maxOutputTokens?: number | undefined;
+  additionalChatOptions?: OpenAIResponsesChatOptions | undefined;
+  reasoningEffort?: "low" | "medium" | "high" | undefined;
+  apiKey?: string | undefined;
+  baseURL?: string | undefined;
+  maxRetries: number;
+  timeout?: number;
+  additionalSessionOptions?:
+    | undefined
+    | Omit<Partial<OpenAIClientOptions>, "apiKey" | "maxRetries" | "timeout">;
+  lazySession: () => Promise<LLMInstance>;
+  #session: Promise<LLMInstance> | null = null;
+
+  trackPreviousResponses: boolean;
+  store: boolean;
+  user: string;
+  callMetadata: StoredValue;
+  builtInTools: OpenAILLM.Responses.Tool[] | null;
+  strict: boolean;
+  include: OpenAILLM.Responses.ResponseIncludable[] | null;
+  instructions: string;
+  previousResponseId: string | null;
+  truncation: "auto" | "disabled" | null;
+
+  constructor(
+    init?: Omit<Partial<OpenAIResponses>, "session"> & {
+      session?: LLMInstance | undefined;
+      azure?: AzureClientOptions;
+    },
+  ) {
+    super();
+    this.model = init?.model ?? "gpt-4o";
+    this.temperature = init?.temperature ?? 0.1;
+
+    this.reasoningEffort = isReasoningModel(this.model)
+      ? init?.reasoningEffort
+      : undefined;
+
+    this.topP = init?.topP ?? 1;
+    this.maxOutputTokens = init?.maxOutputTokens ?? undefined;
+    this.maxRetries = init?.maxRetries ?? 10;
+
+    this.timeout = init?.timeout ?? 60 * 1000;
+
+    this.apiKey =
+      init?.session?.apiKey ?? init?.apiKey ?? getEnv("OPENAI_API_KEY");
+    this.baseURL =
+      init?.session?.baseURL ?? init?.baseURL ?? getEnv("OPENAI_BASE_URL");
+
+    this.additionalSessionOptions = init?.additionalSessionOptions;
+    this.additionalChatOptions = init?.additionalChatOptions;
+
+    this.trackPreviousResponses = init?.trackPreviousResponses ?? false;
+    this.builtInTools = init?.builtInTools ?? null;
+    this.store = init?.store ?? false;
+    this.user = init?.user ?? "";
+    this.callMetadata = init?.callMetadata ?? {};
+    this.strict = init?.strict ?? false;
+    this.include = init?.include ?? null;
+    this.instructions = init?.instructions ?? "";
+    this.previousResponseId = init?.previousResponseId ?? null;
+    this.truncation = init?.truncation ?? null;
+
+    if (init?.azure || shouldUseAzure()) {
+      const azureConfig = {
+        ...getAzureConfigFromEnv({
+          model: getAzureModel(this.model),
+        }),
+        ...init?.azure,
+      };
+
+      this.lazySession = async () =>
+        init?.session ??
+        import("openai").then(({ AzureOpenAI }) => {
+          AzureOpenAI = AzureOpenAIWithUserAgent(AzureOpenAI);
+
+          return new AzureOpenAI({
+            maxRetries: this.maxRetries,
+            timeout: this.timeout!,
+            ...this.additionalSessionOptions,
+            ...azureConfig,
+          });
+        });
+    } else {
+      this.lazySession = async () =>
+        init?.session ??
+        import("openai").then(({ OpenAI }) => {
+          return new OpenAI({
+            apiKey: this.apiKey,
+            baseURL: this.baseURL,
+            maxRetries: this.maxRetries,
+            timeout: this.timeout!,
+            ...this.additionalSessionOptions,
+          });
+        });
+    }
+  }
+
+  get session() {
+    if (!this.#session) {
+      this.#session = this.lazySession();
+    }
+    return this.#session;
+  }
+
+  get supportToolCall() {
+    return isFunctionCallingModel(this);
+  }
+
+  get metadata(): LLMMetadata & OpenAIAdditionalMetadata {
+    const contextWindow =
+      ALL_AVAILABLE_OPENAI_MODELS[
+        this.model as keyof typeof ALL_AVAILABLE_OPENAI_MODELS
+      ]?.contextWindow ?? 1024;
+
+    return {
+      model: this.model,
+      temperature: this.temperature,
+      topP: this.topP,
+      maxTokens: this.maxOutputTokens,
+      contextWindow,
+      tokenizer: Tokenizers.CL100K_BASE,
+      structuredOutput: true,
+    };
+  }
+
+  private createInitialMessage(): ChatMessage<ToolCallLLMMessageOptions> {
+    return {
+      role: "assistant",
+      content: "",
+    };
+  }
+
+  private createInitialOptions(): ResponsesAdditionalOptions {
+    return {
+      built_in_tool_calls: [],
+    };
+  }
+
+  private isBuiltInToolCall(
+    item: OpenAILLM.Responses.ResponseOutputItem,
+  ): item is
+    | OpenAILLM.Responses.ResponseFileSearchToolCall
+    | OpenAILLM.Responses.ResponseComputerToolCall
+    | OpenAILLM.Responses.ResponseFunctionWebSearch {
+    return ["file_search_call", "computer_call", "web_search_call"].includes(
+      item.type,
+    );
+  }
+
+  private isReasoning(item: OpenAILLM.Responses.ResponseOutputItem) {
+    return item.type === "reasoning";
+  }
+
+  private isMessageBlock(item: OpenAILLM.Responses.ResponseOutputItem) {
+    return item.type === "message";
+  }
+
+  private isFunctionCall(item: OpenAILLM.Responses.ResponseOutputItem) {
+    return item.type === "function_call";
+  }
+
+  private isResponseCreatedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseCreatedEvent {
+    return event.type === "response.created";
+  }
+
+  private isResponseOutputItemAddedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseOutputItemAddedEvent {
+    return event.type === "response.output_item.added";
+  }
+
+  private isToolCallEvent(
+    event: OpenAILLM.Responses.ResponseOutputItemAddedEvent,
+  ): event is OpenAILLM.Responses.ResponseOutputItemAddedEvent & {
+    item: OpenAILLM.Responses.ResponseFunctionToolCall;
+  } {
+    return event.item.type === "function_call";
+  }
+
+  private isResponseTextDeltaEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseTextDeltaEvent {
+    return event.type === "response.output_text.delta";
+  }
+
+  private isResponseFunctionCallArgumentsDeltaEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseFunctionCallArgumentsDeltaEvent {
+    return event.type === "response.function_call_arguments.delta";
+  }
+
+  private isResponseFunctionCallDoneEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseFunctionCallArgumentsDoneEvent {
+    return event.type === "response.function_call_arguments.done";
+  }
+
+  private isResponseOutputTextAnnotationAddedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseTextAnnotationDeltaEvent {
+    return event.type === "response.output_text.annotation.added";
+  }
+
+  private isResponseFileSearchCallCompletedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseFileSearchCallCompletedEvent {
+    return event.type === "response.file_search_call.completed";
+  }
+
+  private isResponseWebSearchCallCompletedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseWebSearchCallCompletedEvent {
+    return event.type === "response.web_search_call.completed";
+  }
+
+  private isResponseCompletedEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+  ): event is OpenAILLM.Responses.ResponseCompletedEvent {
+    return event.type === "response.completed";
+  }
+
+  private isTextPresent(
+    part:
+      | OpenAILLM.Responses.ResponseOutputText
+      | OpenAILLM.Responses.ResponseOutputRefusal,
+  ) {
+    return "text" in part;
+  }
+
+  private isRefusalPresent(
+    part:
+      | OpenAILLM.Responses.ResponseOutputText
+      | OpenAILLM.Responses.ResponseOutputRefusal,
+  ) {
+    return "refusal" in part;
+  }
+
+  private isAnnotationPresent(
+    part:
+      | OpenAILLM.Responses.ResponseOutputText
+      | OpenAILLM.Responses.ResponseOutputRefusal,
+  ) {
+    return "annotations" in part;
+  }
+
+  private handleResponseOutputMessage(
+    item: OpenAILLM.Responses.ResponseOutputMessage,
+    options: ResponsesAdditionalOptions,
+  ): string {
+    let outputContent = "";
+    for (const part of item.content) {
+      if (this.isTextPresent(part)) {
+        outputContent += part.text;
+      }
+      if (this.isAnnotationPresent(part)) {
+        options.annotations = part.annotations;
+      }
+      if (this.isRefusalPresent(part)) {
+        options.refusal = part.refusal;
+      }
+    }
+    return outputContent;
+  }
+
+  private extractToolCalls(
+    response: OpenAILLM.Responses.ResponseOutputItem[],
+  ): PartialToolCall[] {
+    return response
+      .filter((item): item is OpenAILLM.Responses.ResponseFunctionToolCall =>
+        this.isFunctionCall(item),
+      )
+      .map((item) => {
+        return {
+          name: item.name,
+          id: item.call_id,
+          input: item.arguments,
+        };
+      });
+  }
+
+  private parseResponseOutput(
+    response: OpenAILLM.Responses.ResponseOutputItem[],
+  ): ChatMessage<ToolCallLLMMessageOptions> {
+    const message = this.createInitialMessage();
+    const options = this.createInitialOptions();
+    const toolCall = this.extractToolCalls(response);
+
+    for (const item of response) {
+      if (this.isMessageBlock(item)) {
+        const outputContent = this.handleResponseOutputMessage(item, options);
+        message.content = outputContent;
+      } else if (this.isBuiltInToolCall(item)) {
+        options.built_in_tool_calls.push(item);
+      } else if (this.isReasoning(item)) {
+        options.reasoning = item;
+      }
+    }
+    message.options = {
+      ...options,
+      toolCall: toolCall,
+    };
+
+    return message;
+  }
+
+  private processStreamEvent(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+    streamState: StreamState,
+  ) {
+    switch (true) {
+      case this.isResponseCreatedEvent(event):
+        this.handleResponseCreatedEvent(event, streamState);
+        break;
+      case this.isResponseOutputItemAddedEvent(event):
+        this.handleOutputItemAddedEvent(event, streamState);
+        break;
+      case this.isResponseTextDeltaEvent(event):
+        this.handleTextDeltaEvent(event, streamState);
+        break;
+      case this.isResponseFunctionCallArgumentsDeltaEvent(event):
+        this.handleFunctionCallArgumentsDeltaEvent(event, streamState);
+        break;
+      case this.isResponseFunctionCallDoneEvent(event):
+        this.handleFunctionCallArgumentsDoneEvent(event, streamState);
+        break;
+      case this.isResponseOutputTextAnnotationAddedEvent(event):
+        this.handleOutputTextAnnotationAddedEvent(event, streamState);
+        break;
+      case this.isResponseFileSearchCallCompletedEvent(event):
+      case this.isResponseWebSearchCallCompletedEvent(event):
+        this.handleBuiltInToolCallCompletedEvent(event, streamState);
+        break;
+      case this.isResponseCompletedEvent(event):
+        this.handleCompletedEvent(event, streamState);
+        break;
+    }
+  }
+
+  private handleResponseCreatedEvent(
+    event: OpenAILLM.Responses.ResponseCreatedEvent,
+    streamState: StreamState,
+  ) {
+    if (this.trackPreviousResponses) {
+      streamState.previousResponseId = event.response.id;
+    }
+  }
+
+  private handleOutputItemAddedEvent(
+    event: OpenAILLM.Responses.ResponseOutputItemAddedEvent,
+    streamState: StreamState,
+  ) {
+    if (this.isToolCallEvent(event)) {
+      streamState.currentToolCall = {
+        name: event.item.name,
+        id: event.item.call_id,
+        input: event.item.arguments,
+      };
+    }
+  }
+
+  private handleTextDeltaEvent(
+    event: OpenAILLM.Responses.ResponseTextDeltaEvent,
+    streamState: StreamState,
+  ) {
+    streamState.delta = event.delta;
+  }
+
+  private handleFunctionCallArgumentsDeltaEvent(
+    event: OpenAILLM.Responses.ResponseFunctionCallArgumentsDeltaEvent,
+    streamState: StreamState,
+  ) {
+    if (streamState.currentToolCall) {
+      streamState.currentToolCall!.input += event.delta;
+    }
+  }
+
+  private handleFunctionCallArgumentsDoneEvent(
+    event: OpenAILLM.Responses.ResponseFunctionCallArgumentsDoneEvent,
+    streamState: StreamState,
+  ) {
+    if (streamState.currentToolCall) {
+      streamState.currentToolCall.input = event.arguments;
+    }
+
+    streamState.shouldEmitToolCall = {
+      ...streamState.currentToolCall!,
+      input: JSON.parse(streamState.currentToolCall!.input),
+    };
+  }
+
+  private handleOutputTextAnnotationAddedEvent(
+    event: OpenAILLM.Responses.ResponseTextAnnotationDeltaEvent,
+    streamState: StreamState,
+  ) {
+    if (!streamState.options.annotations) {
+      streamState.options.annotations = [];
+    }
+    streamState.options.annotations.push(event.annotation);
+  }
+
+  private handleBuiltInToolCallCompletedEvent(
+    event:
+      | OpenAILLM.Responses.ResponseFileSearchCallCompletedEvent
+      | OpenAILLM.Responses.ResponseWebSearchCallCompletedEvent,
+    streamState: StreamState,
+  ) {
+    streamState.options.built_in_tool_calls.push(event);
+  }
+
+  private handleCompletedEvent(
+    event: OpenAILLM.Responses.ResponseCompletedEvent,
+    streamState: StreamState,
+  ) {
+    if (event.response.usage) {
+      streamState.options.usage = event.response.usage;
+    }
+  }
+
+  private createBaseRequestParams(
+    messages: ChatMessage<ToolCallLLMMessageOptions>[],
+    tools: BaseTool[] | undefined,
+    additionalChatOptions: OpenAIResponsesChatOptions | undefined,
+  ) {
+    const baseRequestParams = <OpenAILLM.Responses.ResponseCreateParams>{
+      model: this.model,
+      include: this.include,
+      input: this.toOpenAIResponseMessages(messages),
+      tools: this.builtInTools ? [...this.builtInTools] : [],
+      instructions: this.instructions,
+      max_output_tokens: this.maxOutputTokens,
+      previous_response_id: this.previousResponseId,
+      store: this.store,
+      metadata: this.callMetadata,
+      top_p: this.topP,
+      truncation: this.truncation,
+      user: this.user,
+      ...Object.assign({}, this.additionalChatOptions, additionalChatOptions),
+    };
+
+    if (tools?.length) {
+      if (!baseRequestParams.tools) {
+        baseRequestParams.tools = [];
+      }
+      baseRequestParams.tools.push(
+        ...tools.map(this.toResponsesTool.bind(this)),
+      );
+    }
+
+    return baseRequestParams;
+  }
+
+  chat(
+    params: LLMChatParamsStreaming<
+      OpenAIResponsesChatOptions,
+      ToolCallLLMMessageOptions
+    >,
+  ): Promise<AsyncIterable<ChatResponseChunk<ToolCallLLMMessageOptions>>>;
+  chat(
+    params: LLMChatParamsNonStreaming<
+      OpenAIResponsesChatOptions,
+      ToolCallLLMMessageOptions
+    >,
+  ): Promise<ChatResponse<ToolCallLLMMessageOptions>>;
+  async chat(
+    params:
+      | LLMChatParamsNonStreaming<
+          OpenAIResponsesChatOptions,
+          ToolCallLLMMessageOptions
+        >
+      | LLMChatParamsStreaming<
+          OpenAIResponsesChatOptions,
+          ToolCallLLMMessageOptions
+        >,
+  ): Promise<
+    | ChatResponse<ToolCallLLMMessageOptions>
+    | AsyncIterable<ChatResponseChunk<ToolCallLLMMessageOptions>>
+  > {
+    const { messages, stream, tools, responseFormat, additionalChatOptions } =
+      params;
+    const baseRequestParams = this.createBaseRequestParams(
+      messages,
+      tools,
+      additionalChatOptions,
+    );
+
+    if (
+      Array.isArray(baseRequestParams.tools) &&
+      baseRequestParams.tools.length === 0
+    ) {
+      // remove empty tools array to avoid OpenAI error
+      delete baseRequestParams.tools;
+    }
+
+    if (!isTemperatureSupported(baseRequestParams.model))
+      delete baseRequestParams.temperature;
+
+    if (stream) {
+      return this.streamChat(baseRequestParams);
+    }
+
+    const response = await (
+      await this.session
+    ).responses.create({
+      ...baseRequestParams,
+      stream: false,
+    });
+
+    const message = this.parseResponseOutput(response.output);
+    return {
+      raw: response,
+      message,
+    };
+  }
+
+  private initalizeStreamState(): StreamState {
+    return {
+      delta: "",
+      currentToolCall: null,
+      shouldEmitToolCall: null,
+      options: this.createInitialOptions(),
+      previousResponseId: this.previousResponseId,
+    };
+  }
+
+  private createResponseChunk(
+    event: OpenAILLM.Responses.ResponseStreamEvent,
+    state: StreamState,
+  ): ChatResponseChunk<ToolCallLLMMessageOptions> {
+    return {
+      raw: event,
+      delta: state.delta,
+      options: state.shouldEmitToolCall
+        ? { toolCall: [state.shouldEmitToolCall] }
+        : state.currentToolCall
+          ? { toolCall: [state.currentToolCall] }
+          : {},
+    };
+  }
+
+  @wrapEventCaller
+  protected async *streamChat(
+    baseRequestParams: OpenAILLM.Responses.ResponseCreateParams,
+  ): AsyncIterable<ChatResponseChunk<ToolCallLLMMessageOptions>> {
+    const streamState = this.initalizeStreamState();
+
+    const stream = await (
+      await this.session
+    ).responses.create({
+      ...baseRequestParams,
+      stream: true,
+    });
+
+    for await (const event of stream) {
+      this.processStreamEvent(event, streamState);
+      this.handlePreviousResponseId(streamState);
+      yield this.createResponseChunk(event, streamState);
+    }
+  }
+
+  toOpenAIResponsesRole(messageType: MessageType): OpenAIResponsesRole {
+    switch (messageType) {
+      case "user":
+        return "user";
+      case "assistant":
+        return "assistant";
+      case "system":
+        return "system";
+      case "developer":
+        return "developer";
+      default:
+        return "user";
+    }
+  }
+
+  private isToolResultPresent(
+    options: ToolCallLLMMessageOptions,
+  ): options is ToolResultOptions {
+    return "toolResult" in options;
+  }
+
+  private isToolCallPresent(
+    options: ToolCallLLMMessageOptions,
+  ): options is ToolCallOptions {
+    return "toolCall" in options;
+  }
+
+  private isUserMessage(message: ChatMessage<ToolCallLLMMessageOptions>) {
+    return message.role === "user";
+  }
+
+  private handlePreviousResponseId(streamState: StreamState) {
+    if (this.trackPreviousResponses) {
+      if (streamState.previousResponseId != this.previousResponseId) {
+        this.previousResponseId = streamState.previousResponseId;
+      }
+    }
+  }
+
+  private convertToOpenAIToolCallResult(
+    options: ToolResultOptions,
+    content: MessageContent,
+  ) {
+    return {
+      type: "function_call_output",
+      call_id: options.toolResult.id,
+      output: extractText(content),
+    } satisfies OpenAILLM.Responses.ResponseInputItem.FunctionCallOutput;
+  }
+
+  private convertToOpenAIToolCalls(options: ToolCallOptions) {
+    return options.toolCall.map((toolCall) => {
+      return {
+        type: "function_call",
+        call_id: toolCall.id,
+        name: toolCall.name,
+        arguments:
+          typeof toolCall.input === "string"
+            ? toolCall.input
+            : JSON.stringify(toolCall.input),
+      };
+    }) satisfies OpenAILLM.Responses.ResponseFunctionToolCall[];
+  }
+
+  private processMessageContent(
+    content: MessageContent,
+  ): ResponseMessageContent {
+    if (!Array.isArray(content)) {
+      return content;
+    }
+
+    return content.map((item) => {
+      if (item.type === "text") {
+        return {
+          type: "input_text",
+          text: item.text,
+        };
+      }
+      if (item.type === "image_url") {
+        return {
+          type: "input_image",
+          image_url: item.image_url.url,
+          detail: item.detail || "auto",
+        };
+      }
+      throw new Error("Unsupported content type");
+    });
+  }
+
+  private convertToOpenAIUserMessage(
+    message: ChatMessage<ToolCallLLMMessageOptions>,
+  ) {
+    const messageContent = this.processMessageContent(message.content);
+    return {
+      role: "user",
+      content: messageContent,
+    } satisfies OpenAILLM.Responses.EasyInputMessage;
+  }
+
+  private defaultOpenAIResponseMessage(
+    message: ChatMessage<ToolCallLLMMessageOptions>,
+  ) {
+    const response: OpenAILLM.Responses.ResponseInputItem = {
+      role: this.toOpenAIResponsesRole(message.role),
+      content: extractText(message.content),
+    };
+    return response;
+  }
+
+  toOpenAIResponseMessage(
+    message: ChatMessage<ToolCallLLMMessageOptions>,
+  ):
+    | OpenAILLM.Responses.ResponseInputItem
+    | OpenAILLM.Responses.ResponseInputItem[] {
+    const options = message.options ?? {};
+
+    if (this.isToolResultPresent(options)) {
+      return this.convertToOpenAIToolCallResult(options, message.content);
+    } else if (this.isToolCallPresent(options)) {
+      return this.convertToOpenAIToolCalls(options);
+    } else if (this.isUserMessage(message)) {
+      return this.convertToOpenAIUserMessage(message);
+    }
+
+    return this.defaultOpenAIResponseMessage(message);
+  }
+
+  toOpenAIResponseMessages(
+    messages: ChatMessage<ToolCallLLMMessageOptions>[],
+  ): OpenAILLM.Responses.ResponseInput {
+    const finalMessages: OpenAILLM.Responses.ResponseInputItem[] = [];
+    for (const message of messages) {
+      const processedMessage = this.toOpenAIResponseMessage(message);
+      if (Array.isArray(processedMessage)) {
+        finalMessages.push(...processedMessage);
+      } else {
+        finalMessages.push(processedMessage);
+      }
+    }
+    return finalMessages;
+  }
+
+  toResponsesTool(tool: BaseTool): OpenAILLM.Responses.Tool {
+    return {
+      type: "function",
+      name: tool.metadata.name,
+      description: tool.metadata.description,
+      parameters: tool.metadata.parameters ?? {},
+      strict: this.strict,
+    };
+  }
+}
+
+/**
+ * Convenience function to create a new OpenAI instance.
+ * @param init - Optional initialization parameters for the OpenAI instance.
+ * @returns A new OpenAI instance.
+ */
+export const openaiResponses = (
+  init?: ConstructorParameters<typeof OpenAIResponses>[0],
+) => new OpenAIResponses(init);
@@ -0,0 +1,213 @@
+import type { LLM, PartialToolCall } from "@llamaindex/core/llms";
+import { AzureOpenAI as AzureOpenAILLM, OpenAI as OpenAILLM } from "openai";
+import type { ChatModel } from "openai/resources.mjs";
+import { OpenAI } from "./llm";
+
+export const GPT4_MODELS = {
+  "chatgpt-4o-latest": {
+    contextWindow: 128000,
+  },
+  "gpt-4.5-preview": { contextWindow: 128000 },
+  "gpt-4.5-preview-2025-02-27": { contextWindow: 128000 },
+  "gpt-4": { contextWindow: 8192 },
+  "gpt-4-32k": { contextWindow: 32768 },
+  "gpt-4-32k-0613": { contextWindow: 32768 },
+  "gpt-4-turbo": { contextWindow: 128000 },
+  "gpt-4-turbo-preview": { contextWindow: 128000 },
+  "gpt-4-1106-preview": { contextWindow: 128000 },
+  "gpt-4-0125-preview": { contextWindow: 128000 },
+  "gpt-4-vision-preview": { contextWindow: 128000 },
+  "gpt-4o": { contextWindow: 128000 },
+  "gpt-4o-2024-05-13": { contextWindow: 128000 },
+  "gpt-4o-mini": { contextWindow: 128000 },
+  "gpt-4o-mini-2024-07-18": { contextWindow: 128000 },
+  "gpt-4o-2024-08-06": { contextWindow: 128000 },
+  "gpt-4o-2024-09-14": { contextWindow: 128000 },
+  "gpt-4o-2024-10-14": { contextWindow: 128000 },
+  "gpt-4-0613": { contextWindow: 128000 },
+  "gpt-4-turbo-2024-04-09": { contextWindow: 128000 },
+  "gpt-4-0314": { contextWindow: 128000 },
+  "gpt-4-32k-0314": { contextWindow: 32768 },
+  "gpt-4o-realtime-preview": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-realtime-preview-2024-10-01": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-audio-preview": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-audio-preview-2024-10-01": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-2024-11-20": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-audio-preview-2024-12-17": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-mini-audio-preview": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-mini-audio-preview-2024-12-17": {
+    contextWindow: 128000,
+  },
+  "gpt-4o-search-preview": { contextWindow: 128000 },
+  "gpt-4o-mini-search-preview": { contextWindow: 128000 },
+  "gpt-4o-search-preview-2025-03-11": { contextWindow: 128000 },
+  "gpt-4o-mini-search-preview-2025-03-11": { contextWindow: 128000 },
+};
+
+// NOTE we don't currently support gpt-3.5-turbo-instruct and don't plan to in the near future
+export const GPT35_MODELS = {
+  "gpt-3.5-turbo": { contextWindow: 16385 },
+  "gpt-3.5-turbo-0613": { contextWindow: 4096 },
+  "gpt-3.5-turbo-16k": { contextWindow: 16385 },
+  "gpt-3.5-turbo-16k-0613": { contextWindow: 16385 },
+  "gpt-3.5-turbo-1106": { contextWindow: 16385 },
+  "gpt-3.5-turbo-0125": { contextWindow: 16385 },
+  "gpt-3.5-turbo-0301": { contextWindow: 16385 },
+};
+
+export const O1_MODELS = {
+  "o1-preview": {
+    contextWindow: 128000,
+  },
+  "o1-preview-2024-09-12": {
+    contextWindow: 128000,
+  },
+  "o1-mini": {
+    contextWindow: 128000,
+  },
+  "o1-mini-2024-09-12": {
+    contextWindow: 128000,
+  },
+  o1: {
+    contextWindow: 128000,
+  },
+  "o1-2024-12-17": {
+    contextWindow: 128000,
+  },
+};
+
+export const O3_MODELS = {
+  "o3-mini": {
+    contextWindow: 200000,
+  },
+  "o3-mini-2025-01-31": {
+    contextWindow: 200000,
+  },
+};
+
+/**
+ * We currently support GPT-3.5 and GPT-4 models
+ */
+export const ALL_AVAILABLE_OPENAI_MODELS = {
+  ...GPT4_MODELS,
+  ...GPT35_MODELS,
+  ...O1_MODELS,
+  ...O3_MODELS,
+} satisfies Record<ChatModel, { contextWindow: number }>;
+
+export function isFunctionCallingModel(llm: LLM): llm is OpenAI {
+  let model: string;
+  if (llm instanceof OpenAI) {
+    model = llm.model;
+  } else if ("model" in llm && typeof llm.model === "string") {
+    model = llm.model;
+  } else {
+    return false;
+  }
+  const isChatModel = Object.keys(ALL_AVAILABLE_OPENAI_MODELS).includes(model);
+  const isOld = model.includes("0314") || model.includes("0301");
+  const isO1 = model.startsWith("o1");
+  return isChatModel && !isOld && !isO1;
+}
+
+export function isReasoningModel(model: ChatModel | string): boolean {
+  const isO1 = model.startsWith("o1");
+  const isO3 = model.startsWith("o3");
+  return isO1 || isO3;
+}
+
+export function isTemperatureSupported(model: ChatModel | string): boolean {
+  return !model.startsWith("o3");
+}
+
+export type OpenAIAdditionalMetadata = object;
+
+export type OpenAIAdditionalChatOptions = Omit<
+  Partial<OpenAILLM.Chat.ChatCompletionCreateParams>,
+  | "max_tokens"
+  | "messages"
+  | "model"
+  | "temperature"
+  | "reasoning_effort"
+  | "top_p"
+  | "stream"
+  | "tools"
+  | "toolChoice"
+>;
+
+export type LLMInstance = Pick<
+  AzureOpenAILLM | OpenAILLM,
+  "chat" | "apiKey" | "baseURL" | "responses"
+>;
+
+export type OpenAIResponsesChatOptions = Omit<
+  Partial<OpenAILLM.Responses.ResponseCreateParams>,
+  | "model"
+  | "input"
+  | "stream"
+  | "tools"
+  | "toolChoice"
+  | "temperature"
+  | "reasoning_effort"
+  | "top_p"
+  | "max_output_tokens"
+  | "include"
+>;
+
+export type ResponsesAdditionalOptions = {
+  built_in_tool_calls: Array<
+    | OpenAILLM.Responses.ResponseFileSearchToolCall
+    | OpenAILLM.Responses.ResponseComputerToolCall
+    | OpenAILLM.Responses.ResponseFunctionWebSearch
+    | OpenAILLM.Responses.ResponseFileSearchCallCompletedEvent
+    | OpenAILLM.Responses.ResponseWebSearchCallCompletedEvent
+  >;
+  annotations?: Array<
+    | OpenAILLM.Responses.ResponseOutputText.FileCitation
+    | OpenAILLM.Responses.ResponseOutputText.URLCitation
+    | OpenAILLM.Responses.ResponseOutputText.FilePath
+  >;
+  refusal?: string;
+  reasoning?: OpenAILLM.Responses.ResponseReasoningItem;
+  usage?: OpenAILLM.Responses.ResponseUsage;
+};
+
+export type StreamState = {
+  delta: string;
+  currentToolCall: PartialToolCall | null;
+  shouldEmitToolCall: PartialToolCall | null;
+  options: ResponsesAdditionalOptions;
+  previousResponseId: string | null;
+};
+
+export type ResponsesMessageContentTextDetail = {
+  type: "input_text";
+  text: string;
+};
+
+export type ResponsesMessageContentImageDetail = {
+  type: "input_image";
+  image_url: string;
+  detail: "high" | "low" | "auto";
+};
+export type ResponsesMessageContentDetail =
+  | ResponsesMessageContentTextDetail
+  | ResponsesMessageContentImageDetail;
+
+export type ResponseMessageContent = string | ResponsesMessageContentDetail[];
+
+export type OpenAIResponsesRole = "user" | "assistant" | "system" | "developer";
@@ -0,0 +1,223 @@
+import type { BaseTool, ToolCallOptions } from "@llamaindex/core/llms";
+import { describe, expect, it } from "vitest";
+import { OpenAIResponses } from "../src/responses";
+
+const API_KEY = process.env.MY_OPENAI_API_KEY;
+
+describe("OpenAIResponses Integration Tests", () => {
+  if (!API_KEY) {
+    describe.skip("OpenAI API key not found skipping tests");
+    return;
+  }
+
+  const llm = new OpenAIResponses({
+    model: "gpt-4o",
+    apiKey: API_KEY,
+  });
+
+  it("should handle basic text chat", async () => {
+    const response = await llm.chat({
+      messages: [
+        {
+          role: "user",
+          content: "What is 2+2? Answer in one word.",
+        },
+      ],
+    });
+
+    expect(response.message.content).toBe("Four.");
+    expect(response.message.role).toBe("assistant");
+  });
+
+  it("should handle image analysis", async () => {
+    const response = await llm.chat({
+      messages: [
+        {
+          role: "user",
+          content: [
+            {
+              type: "text",
+              text: "What's in this image? Describe in one sentence.",
+            },
+            {
+              type: "image_url",
+              image_url: {
+                url: "https://storage.googleapis.com/cloud-samples-data/vision/face/faces.jpeg",
+              },
+            },
+          ],
+        },
+      ],
+    });
+    expect(response.raw).toHaveProperty("id");
+  });
+
+  it("should handle function calls", async () => {
+    const weatherTool: BaseTool = {
+      metadata: {
+        name: "get_weather",
+        description: "Get current weather for a location",
+        parameters: {
+          type: "object",
+          properties: {
+            location: {
+              type: "string",
+              description: "City name",
+            },
+          },
+          required: ["location"],
+        },
+      },
+    };
+
+    const response = await llm.chat({
+      messages: [
+        {
+          role: "user",
+          content: "What's the weather in London?",
+        },
+      ],
+      tools: [weatherTool],
+    });
+
+    expect(
+      (response.message.options as ToolCallOptions)?.toolCall,
+    ).toBeDefined();
+    const toolCall = (response.message.options as ToolCallOptions)
+      ?.toolCall?.[0];
+    expect(toolCall?.name).toBe("get_weather");
+    expect(
+      typeof toolCall?.input === "string" &&
+        JSON.parse(toolCall?.input || "{}"),
+    ).toHaveProperty("location", "London");
+  });
+});
+
+describe("OpenAIResponses Unit Tests", () => {
+  // Testing utility functions
+  describe("processMessageContent", () => {
+    const llm = new OpenAIResponses({
+      model: "gpt-4o",
+      apiKey: "test",
+    });
+
+    it("should handle non-array content (string)", () => {
+      const content = "Hello world";
+      // @ts-expect-error accessing private method
+      const result = llm.processMessageContent(content);
+      expect(result).toBe(content);
+    });
+
+    it("should process text content", () => {
+      const content = [
+        {
+          type: "text",
+          text: "Hello world",
+        },
+      ];
+      // @ts-expect-error accessing private method
+      const result = llm.processMessageContent(content);
+      expect(result[0]).toEqual({
+        type: "input_text",
+        text: "Hello world",
+      });
+    });
+
+    it("should process image content with default detail", () => {
+      const content = [
+        {
+          type: "image_url",
+          image_url: { url: "https://example.com/image.jpg" },
+        },
+      ];
+      // @ts-expect-error accessing private method
+      const result = llm.processMessageContent(content);
+      expect(result[0]).toEqual({
+        type: "input_image",
+        image_url: "https://example.com/image.jpg",
+        detail: "auto",
+      });
+    });
+
+    it("should process image content with specified detail", () => {
+      const content = [
+        {
+          type: "image_url",
+          image_url: { url: "https://example.com/image.jpg" },
+          detail: "high",
+        },
+      ];
+      // @ts-expect-error accessing private method
+      const result = llm.processMessageContent(content);
+      expect(result[0]).toEqual({
+        type: "input_image",
+        image_url: "https://example.com/image.jpg",
+        detail: "high",
+      });
+    });
+
+    it("should process mixed content", () => {
+      const content = [
+        {
+          type: "text",
+          text: "What's in this image?",
+        },
+        {
+          type: "image_url",
+          image_url: { url: "https://example.com/image.jpg" },
+        },
+      ];
+      // @ts-expect-error accessing private method
+      const result = llm.processMessageContent(content);
+      expect(result).toEqual([
+        {
+          type: "input_text",
+          text: "What's in this image?",
+        },
+        {
+          type: "input_image",
+          image_url: "https://example.com/image.jpg",
+          detail: "auto",
+        },
+      ]);
+    });
+  });
+
+  describe("isResponseCreatedEvent", () => {
+    const llm = new OpenAIResponses({
+      model: "gpt-4o",
+      apiKey: "test",
+    });
+
+    it("should identify response created events", () => {
+      const event = { type: "response.created", response_id: "123" };
+      // @ts-expect-error accessing private method
+      expect(llm.isResponseCreatedEvent(event)).toBe(true);
+    });
+
+    it("should reject non-created events", () => {
+      const event = { type: "response.other", response_id: "123" };
+      // @ts-expect-error accessing private method
+      expect(llm.isResponseCreatedEvent(event)).toBe(false);
+    });
+  });
+
+  describe("isFunctionCall", () => {
+    const llm = new OpenAIResponses({
+      model: "gpt-4o",
+      apiKey: "test",
+    });
+
+    it("should identify function calls", () => {
+      const item = { type: "function_call", name: "test", arguments: "{}" };
+      // @ts-expect-error accessing private method
+      expect(llm.isFunctionCall(item)).toBe(true);
+    });
+
+    it("should reject non-function calls", () => {
+      const item = { type: "message", content: "test" };
+      // @ts-expect-error accessing private method
+      expect(llm.isFunctionCall(item)).toBe(false);
+    });
+  });
+});
@@ -1,5 +1,13 @@
 # @llamaindex/perplexity

+## 0.0.5
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/openai@0.3.0
+  - @llamaindex/core@0.6.2
+
 ## 0.0.4

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/perplexity",
  "description": "Perplexity Adapter for LlamaIndex",
-  "version": "0.0.4",
+  "version": "0.0.5",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/portkey-ai

+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.43

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/portkey-ai",
  "description": "Portkey Adapter for LlamaIndex",
-  "version": "0.0.43",
+  "version": "0.0.44",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/replicate

+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.43

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/replicate",
  "description": "Replicate Adapter for LlamaIndex",
-  "version": "0.0.43",
+  "version": "0.0.44",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/astra

+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.15

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/astra",
  "description": "Astra Storage for LlamaIndex",
-  "version": "0.0.15",
+  "version": "0.0.16",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/azure

+## 0.1.11
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.1.10

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/azure",
  "description": "Azure Storage for LlamaIndex",
-  "version": "0.1.10",
+  "version": "0.1.11",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,12 @@
 # @llamaindex/chroma

+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
 ## 0.0.15

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/chroma",
  "description": "Chroma Storage for LlamaIndex",
-  "version": "0.0.15",
+  "version": "0.0.16",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Alex Yang	def2b92686	fix:	2025-04-08 13:34:14 -07:00
Alex Yang	6c75b13154	feat: fix type	2025-04-08 13:22:36 -07:00
Alex Yang	cd07b034bf	feat: workflow-based llama-reader	2025-04-08 12:28:30 -07:00
Alex Yang	739323032b	feat: workflow-based llama-reader	2025-04-07 00:50:16 -07:00
Thuc Pham	3534c373f2	feat: support multi-resolution compatibility (#1816 )	2025-04-05 18:41:39 +07:00
github-actions[bot]	2cbdf71669	Release 0.9.16 (#1811 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-04-04 11:12:27 +02:00
Huu Le	ead657aedd	feat: add MCP tools integration and example usage (#1819 )	2025-04-04 11:03:10 +02:00
Marcus Schiesser	f5e4d098b0	chore: remove gpt-tokenizer (#1815 )	2025-04-03 14:23:31 +02:00
dependabot[bot]	4d97226e50	chore(deps): bump next from 15.2.3 to 15.2.4 (#1812 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2025-04-03 13:57:39 +07:00
Marcus Schiesser	4999df18cc	chore: bump nextjs to "^15.2.3" (#1810 )	2025-04-02 21:24:57 +07:00
github-actions[bot]	9a27b6d94a	Release 0.9.15 (#1807 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-04-02 17:08:47 +07:00
Thuc Pham	8c02684f0f	fix: handle error when streaming workflow (#1808 )	2025-04-02 16:26:01 +07:00
ANKIT VARSHNEY	9c63f3f94e	feat: openai responses api (#1801 )	2025-04-02 16:21:43 +07:00
Thuc Pham	c515a324f6	feat: return raw output for agent toolcall result (#1806 )	2025-04-01 22:20:06 +07:00