We are super excited to make Restate Cloud publicly available today. Durable Execution, observability, and production-grade availability for workflows or AI agents. This release makes very smooth to work with serverless compute platforms like Cloudflare, Vercel, or Deno. And of course, your own infra on AWS, GCP, or Kubernetes.... Signup is free and a few clicks with your GitHub account. A new built-in tutorial let's you run a first durable agent running in seconds. 👇show more

Read all about the new Restate Cloud here: We are also launching a new web site today at There is also a fun weird mini game, in case you need a distraction 🤡 👇

The biggest thing for me personally is, though, the unbelievable thing that this team has pulled off: To create a system that is used by 𝘁𝗶𝗲𝗿 𝟭 𝗯𝗮𝗻𝗸𝘀 and 𝗔𝗜 𝘀𝘁𝗮𝗿𝘁𝘂𝗽𝘀 alike. Self-contained, single binary. Works well in cloud-native and serverless environments, also runs multi-DC. Great latency. Both the foundation for our cloud service, and for the on-prem work we have been doing with users and customers. Check out some of the nice things users say about Restate: Pretty amazing work by the whole @restatedev team!

And stay tuned, there is more coming in the next few days!

@Cloudflare @vercel How would you compare this to Inngest?

There is definitely some overlap with Inngest. With the caveat that I am not an expert on them, I would say that Restate is somewhere in between Temporal and Inngest. Restate has a durable functions programming model that is closer to Inngest: meaning it is a simpler API than Temporal and works well with FaaS and serverless compute platforms like Vercel, Cloudflare, etc. More similar to Temporal is that it can be self-hosted in a distributed scale-out manner, that's why Restate is also used on-prem by banks or companies with high compliance requirements (some even with sophisticated geo-distributed failover) or by users with very high durable task throughput (one user is going to 20k invocations per second). What is different than both: (1) Self-contained single binary architecture. Not just for dev or single node, but Restate contains a sophisticated low-latency replicated log that backs everything. It is extremely powerful let's you start on one node and incrementally scale out, even to multi-region clusters with hierarchical replication topology. Easy to start, and simultaneously lets us support scenarios that are otherwise hard to support if you build on top of a database, while maintaining great latency, and comparatively simple operations. Here is a blog post about the design: (2) Restate's programming model has a few unique parts like virtual objects (transactional stateful entities, think similar to stateful virtual actors) (3) I don't know about Inngest's latency, but our users tell us Restate compares quite favorable to Temporal's latency. Restate has a log-based architecture that keeps latencies pretty low. The picture shows some latencies (pure overhead of the runtime, no network or application logic) I am probably missing a bunch more differences. The principle inspiration to Restate is here, basically a unified command-sourcing log:

@Cloudflare @vercel Hmmm. That helps a bit. A comparison on the website would be super useful to make me consider trying it.

@Cloudflare @vercel Exciting launch! Restate Cloud looks like a game changer for building reliable AI workflows ☁️🤖

@Cloudflare @vercel check dm

For the fun

@Cloudflare @vercel durability, available, resilient, at scale love this guys ~ will try your paas let's launch/spotlight you on whenever it helps