prediction: as ai relies more and more on websearch, seo for ai will bring product placement and then it will be followed by ads for ai (content suggestions), esp. in free ai chatbot tiers. ai chatbots will become more like google search in some of those ways.

I find this research fascinating because it means that there is a path towards smaller models that are strong on tool use and reasoning and that use external knowledge sources (through tools such as web search) instead of baked in knowledge. openai.com/index/why-la...

The previously sensible advice to never use ChatGPT for search needs to be rethought - GPT-5 in thinking mode is shockingly good at running searches now simonwillison.net/2025/Sep/6/r...

At React Universe Conference Wroclaw today - would love to chat with AI SDK users. If you are around and want to give feedback or learn more about the AI SDK send me a DM.

Finally - almost 5 months of work, our biggest release so far. Completely revamped, type-safe AI chat is a game changer imo. I don't know any other framework that has full stack support for this. Hope it will enable you to build the next generation of AI applications.

You can try AI SDK 5 now:

AI SDK 5 prototype: custom UI data parts You can define custom ui data parts and stream updates to them from the server:

AI SDK 5 preview: ChatStore & ChatTransport ChatStore synchronizes chat write operations and manages streaming state. You can use it directly or through framework integrations like useChat. ChatTransport makes backend integrations more flexible, allowing for client-only usage.

AI SDK 5 preview: UI message persistance We recommend storing UI messages, not model messages, if your application has a useChat component. This ensures that the UI state can be correctly restored and helps integrating backends other than streamText (e.g. Langchain).

AI SDK 5 preview: UI message metadata UI messages in AI SDK 5 will have a generic metadata property (instead of specific properties like createdAt). This lets you send and show the message metadata that's important in your application:

UI messages != model messages What you display to the user (ui messages) is different from what you want to send to the LLM (model messages). UI messages contain additional information such as custom app data and metadata. Tools calls might be omitted. etc.

In AI SDK 5, UI messages will be composed of parts. This is my current vision of how UI messages could look like - feedback welcome.

Starting to be excited about AI SDK 5 Our planned timeline: - alpha 1st half of May - beta 2nd half of May - GA in June

AI agents vs workflow runs & run termination: Workflows have a deterministic number of steps after which they will finish with with a solution (of varying quality). Agents run in a loop and it is unclear when and if they will finish with a solution.

one fundamental issue w/ ai agents: the problem of deciding when to stop an agent run is a variant of the halting problem there will always be agent runs for which, given their current state and history, it cannot be known whether they will terminate with an answer or not

Combining the provider registry with custom providers and middleware is extremely powerful for setting up your models in one place:

You can set custom headers, e.g. for auth, when using the MCP SSE transport:

Lots of new features shipped since January, check out our 4.2 release:

The AI SDK now supports LLMs that generate file outputs, first example are images:

lot of people asked for this, and while I can claim ~zero credit I'm very happy it's out there really digging the class-based APIs (`chat = new Chat(...)` etc) that are popping up in the Svelte ecosystem. Classes are great! They feel solid, tangible. I missed them.

AI SDK Svelte 2.0 is here with native Svelte 5 support. Thanks to Elliot and @rich-harris.dev for this huge update!

The AI SDK now supports the OpenAI Responses API:

Sometimes you need to access provider-specific information from the responses that is not directly extracted by the AI SDK. When you use generateText or generateObject, you can now access the full JSON response body and use additional data:

Send your own URL sources to the client: bsky.app/profile/sdk....

Repair the model output when there are parse errors with generateObject:

First step towards citation support:

The granular useChat status messages make it easy to show a loading spinner (when status=submitted):

Huge chatbot template upgrade, check it out:

Valibot is getting close to its 1.0 release. To give you more options, we added support for using Valibot schemas with the AI SDK:

Expo / React native requires some polyfilling to work with the AI SDK:

Consuming the stream fully on the server solves the problem where chats get into broken states because the user closes the tab or the network connection breaks down:

streamText immediately starts streaming and errors are part of the stream to prevent server crashes. Unfortunately that means you can't just capture them with try...catch I've added streamText.onError so you can provide a callback for error logging:

My first AI course is complete... And it's free! It's a 16-lesson tutorial on Vercel's AI SDK. It's the missing piece for working with AI in TypeScript. No bullshit, no hype. Just solid, practical advice. https://www.aihero.dev/vercel-ai-sdk-tutorial

AI SDK 4.1 • Image generation • Non-blocking data streaming • Simplified persistence with useChat • Stream transformation and smoothing • Improvements to tool calling and structured outputs

When you store the chats on the server, you only need to send the message the user has entered and not the full chat:

Improving persistent chats with useChat and streamText further:

AI SDK 4.0.37 We've added chat ID forwarding and a helper function to simplify implementing persistence with useChat and streamText. To learn more, check out our latest guide and minimal example.

New AI SDK Provider: @replicate.com Replicate hosts a large number of image models - you can use all of them in the AI SDK now! Check out how to generate images with Flux:

AI SDK 4.0.26 With the new smoothStream 'line' option, you can stream out full lines with a configurable delay:

streaming + structured outputs + tool calls

Updated o1 support in the AI SDK:

Different LLM providers have different capability strengths (in addition to core reasoning/generation ability): - OpenAI: structured outputs (also w/ tool calls) - Anthrophic: PDF, computer use - Google: PDF, audio/video input, search integration

AI SDK provider: Fireworks AI Using Fireworks AI models with the AI SDK is now even easier with the new Fireworks provider:

AI SDK Google Provider New: Search Grounding To have up-to-date information and give you confidence in the results, language models need to be able to access the web. The Gemini models can now search the web automatically:

When structured data generation fails, you can now access more information for local debugging, logging at scale, or error recovery:

Improve your chatbot UX with this simple one-line change:

AI SDK Community Provider: @OpenRouterAI Use the OpenRouter LLM gateway with the AI SDK:

Wanted to add image generation for a long time. It's now finally here (and the API is a solid improvement over what I had in ModelFusion):

📜 AI SDK Template: AI Facts Transcribe and analyze your speech in real-time with our latest template! Powered by Deepgram, Perplexity, and OpenAI (template link in reply).

🧩 AI SDK Observability Integration: @traceloopdev Observability and evals are crucial when developing AI applications. You can use Traceloop with the AI SDK to monitor and evaluate your LLM calls: sdk.vercel.ai/providers/ob...