fix(openai): Allow o1 to stream a single chunk

libs/langchain-openai/src/chat_models.ts

@@ -1320,7 +1320,6 @@ export class ChatOpenAI<
       stream: true as const,
     };
     let defaultRole: OpenAIRoleEnum | undefined;
-
     const streamIterable = await this.completionWithRetry(params, options);
     let usage: OpenAIClient.Completions.CompletionUsage | undefined;
     for await (const data of streamIterable) {
@@ -1599,6 +1598,24 @@ export class ChatOpenAI<
         );
         generations.push(generation);
       }
+
+      await runManager?.handleLLMNewToken(
+        generations[0].text ?? "",
+        {
+          prompt: usageMetadata.input_tokens,
+          completion: usageMetadata.output_tokens,
+        },
+        undefined,
+        undefined,
+        undefined,
+        {
+          chunk: new ChatGenerationChunk({
+            message: new AIMessageChunk({ ...generations[0].message }),
+            text: generations[0].text ?? "",
+          }),
+        }
+      );
+
       return {
         generations,
         llmOutput: {

libs/langchain-openai/src/tests/chat_models.int.test.ts

@@ -26,6 +26,7 @@
   ChatCompletionMessage,
 } from "openai/resources/index.mjs";
 import { ChatOpenAI } from "../chat_models.js";
+import { RunnableLambda } from "@langchain/core/runnables";
 
 // Save the original value of the 'LANGCHAIN_CALLBACKS_BACKGROUND' environment variable
 const originalBackground = process.env.LANGCHAIN_CALLBACKS_BACKGROUND;
@@ -1285,3 +1286,30 @@
     console.log(chunk);
   }
 });
+
+test.only("Streaming with o1 will yield at least one chunk with content", async () => {
+  const model = new ChatOpenAI({
+    model: "o1",
+  });
+
+  const runnable = RunnableLambda.from(() => model.streamEvents(["user", "What color is the sky?"], {
+    version: "v2",
+  }));
+
+  const result = runnable.streamEvents({}, {
+    version: "v2"
+  });
+
+  let content = "";
+  let numStreamChunks = 0;
+  for await (const chunk of result) {
+    console.log(chunk.event)
+    if (chunk.event === "on_chat_model_stream") {
+      content += chunk.data.chunk.content;
+      numStreamChunks += 1;
+    }
+  }
+
+  expect(content.length).toBeGreaterThan(10);
+  expect(numStreamChunks).toBe(1);
+});