3.0.20 create a breaking change for StreamingTextResponse

The problem

Issue vercel/ai#1316 (closed, 37 comments): ### Description the correct response with ai@3.0.19: [image] the error response with ai@3.0.20: [image] it cause an issue with it: https://github.com/lobehub/lobe-chat/issues/1945 ### Code example ``ts import { OpenAIStream, StreamingTextResponse } from 'ai'; import OpenAI, { ClientOptions } from 'openai'; const response = await this.client.chat.completions.create(postPayload); return new StreamingTextResponse(OpenAIStream(response)); ### Additional context I check the update code, I think it's a breaking change for StreamingTextResponse` : [image] please consider to revert it. Or I have

The fix

StreamingTextResponse breaking change in ai 3.0.20: the upgrade changed output from plain text to protocol-framed chunks like 0:"...". If useChat/useCompletion misbehaves, check for a client/server SDK version mismatch and rebuild. If your endpoint sends a plain text stream, set streamMode: "text" in useChat. With custom server code, pin the ai package version.

Notes

Thread: https://github.com/vercel/ai/issues/1316. Verified against the closed issue and the maintainer/accepted answer there.

In https://github.com/vercel/ai/issues/1316, the reporter asked: Issue vercel/ai#1316 (closed, 37 comments): ### Description the correct response with ai@3.0.19: [image] the error response with ai@3.0.20: [image] it cause an issue with it: https://github.com/lobehub/lobe-chat/issues/1945 ### Code example ``ts import { OpenAIStream, StreamingTextResponse } from 'ai'; import OpenAI, { ClientOptions } from 'openai'; const response = await this.client.chat.completions.create(postPayload); return new StreamingTextResponse(OpenAIStream(response)); ### Additional context I check the update code, I think it's a breaking change for StreamingTextResponse : [image] please consider to revert it. Or I have The verified answer: StreamingTextResponse breaking change in ai 3.0.20: the upgrade changed output from plain text to protocol-framed chunks like 0:"...". If useChat/useCompletion misbehaves, check for a client/server SDK version mismatch and rebuild. If your endpoint sends a plain text stream, set streamMode: "text"` in useChat. With custom server code, pin the ai package version. Source thread: https://github.com/vercel/ai/issues/1316