TANSTACK

We built Start with a client first mindset, but server first performance might be our thing too! 😉 New blog post on how we dramatically improved TanStack Start’s SSR performance: 427 req/s → 2357 req/s 424ms avg → 43ms 6558ms p99 → 928ms 99.96% success → 100% 🔗⬇️🧵

Independent benchmark: @Platformatic put TanStack Start, React Router, and Next.js under load and shared the results. Worth a read if you care about SSR performance beyond toy benchmarks. blog.platformatic.dev/react-ssr-fram…

The cool thing about building Void is that I didn’t need to change a single thing about Vite. Void is a plugin that wraps Cloudflare’s Vite plugin - which in turn builds on top of Vite Environment API. And it just composes with frameworks that ship as a Vite plugin, the smoothest being TanStack Start. “Everything is a Vite plugin” in action.

Introducing the @Tan_Stack Intent Skills Registry Intent may use NPM as it's *distribution* registry (awesome), but finding/browsing skills in a sea of NPM modules definitely deserves it's own UI/service - Mirrored from NPM - Package/Skill Search - History/Changelogs Coming soon: - API/CLI for search/lookup - Cross discovery for non-intent skills 🔗⬇️🧵

Building realtime voice AI is usually a pain: WebRTC, audio streams, tool execution, state 😵 TanStack AI now handles that w/ useRealtimeChat()🎙️ 🔊 Voice in / Voice out 🧠 Tools/State 🖼 Multimodal ⚡ Realtime transcripts Supports OpenAI Realtime & ElevenLabs 🔗⬇🧵

Shipping every tool to your LLM with every request simply does not scale. ❌ TanStack AI now ships with Lazy Tool Discovery! 🔎 Tools marked `lazy: true` will be discovered *only* when needed, which means: ⚡ fewer tokens ⚡ better tool selection ⚡ lower cost 🔗⬇︎🧵

Typo: *****shouldn’t**** require 🤦‍♂️

Adding AI features should require 5 different mental models Generation Hooks in TanStack AI unify everything! 🖼️ useGenerateImage() 🎤 useGenerateSpeech() 📝 useTranscription() 📄 useSummarize() 🎬 useGenerateVideo() Same API, state model and streaming pipeline 🔗⬇︎🧵

We just had an astounding release with @tan_stack AI. Generational functions, realtime support. And we have a follow-up that's going to blow your mind and redefine how you think about the agentic lifecycle.

Your chat() endpoint shouldn’t be 200 lines long. TanStack AI now supports full-lifecycle Middleware!🧩 ⚙️ Config transforms 📦 Tool caching 🛡 Content guards 📊 Token tracking 🚦 Rate limits Composable. Type-safe. Streaming-aware. Robust. Blog 🔗⬇︎🧵

Building agentic apps? Use TanStack Start for full-stack delivery, TanStack AI for type-safe tool calling, and TanStack DB for live incremental queries + optimistic mutations. Add PowerSync for offline-first sync, and your app keeps moving even when the network doesn’t. Build for the PowerSync AI Hackathon and compete for cash prizes plus an hour of @tannerlinsley's time and some amazing @tan_stack swag!

Very cool to see my import protection work in TanStack Start inspire @nuxt_js 🙌 youtu.be/JvatoqX595I?t=…

TanStack Hotkeys now has adapters for React, Preact, Vue, Angular, Solid, Svelte, and another potential one on the way for Lit thanks to community contributions.

Excited to welcome @openrouter as an official TanStack sponsor! 🎉 Easily one of our best adapters: 🌐 300+ models from 60+ providers 🔀 Automatic routing and fallbacks 🔓 No lock-in/rewrites when a better model drops 🎨 Goes way beyond chat w/ Image/video/more 🔗⬇︎🧵