The Vercel Chat SDK now features stream resumption. This makes AI conversations resilient to network hiccups and reloading or sharing a chat mid-generation. This is especially valuable for long responses (e.g.: Deep Research). No proprietary APIs, no sticky load balancing, just Redis pubsub.

If you try that example with ChatGPT, it dies and never recovers, unless you wait for long enough for the generation to have completed and refresh the page again:

Demo: Package: PR:

Want to build a Web3 app without all the overhead costs? Try TransactionKit, a React toolkit that makes building dApps easy. Integrate cutting-edge account abstraction features in just a few clicks. Try it today 👇

@vercel when in @aisdk?

@vercel @aisdk cc @max_leiter

@vercel Wait that’s actually pretty sick.

@vercel Build a full-stack deep research application in 80 lines of code with @ValyuNetwork 👇

@vercel Love the work with aisdk. Thank you!

@vercel @t3dotchat been building a sync engine and this came out

@vercel That's cool. I would love to -fully- customize the components.

@vercel beautiful.