Загрузка видео...
Не удалось загрузить видео
Comfy API is live Point it at a workflow JSON and it handles the rest: → Resolves your custom nodes, models, LoRAs, and Python deps → Pins it all, ComfyUI version included, into a versioned Build → Cuts an immutable release, so what you test is what ships →... show more
4,273,173 просмотров • 1 день назад •via X (Twitter)
Комментарии: 37
This is where Comfy workflows start becoming actual infrastructure.
Deploy your first workflow today:
Learn more with our blog:
Huge….if true And it’s true!
great work team!! this is awesome
It is time.
lol
This is super exciting for builders that need custom nodes that aren't supported on Comfy Cloud!!
Time to try this out myself!
@andrewdsouza take a look: ComfyUI's API turns visual AI workflows into versioned, autoscaling GPU endpoints for AI teams. I can help them reach those buyers.
guess who’s gonna be dropping apps like crazy
♥️
game changer!
Most definitely is!
I think the internal-tool use case is going to be huge here.
Finally. Shipping your workflow right from local desktop is a big game changer!
This could be extremely useful for teams building internal creative tools.
The dependency management part might actually be my favorite feature here.
Very interesting for anyone building AI products on top of Comfy.
The local-to-production gap is getting much smaller here.
The immutable release approach is particularly interesting for production workloads.
Really interesting direction for teams already deep into ComfyUI.
This is what happens when the workflow becomes important enough to deserve proper infrastructure.
Immutable releases for workflows finally solve the dependency hell
The reproducibility here is what stands out to me.
Build once, expose it through an endpoint, and let the application handle the rest.
The versioning piece alone could save teams a lot of debugging.
This opens up a lot of possibilities for customer-facing AI features.
The ability to keep the graph behind the product is seriously useful.
This feels like an important step toward production-grade visual AI workflows.
The ability to scale workers based on demand makes this much more practical.
Turning a visual workflow into something an app can call is powerful.
Models + LoRAs + nodes + dependencies in one Build makes a lot of sense.
This is exactly the kind of boring infrastructure work that developers don't want to spend weeks maintaining.
This should make repeatable creative automation much easier.
Turning a graph into something an app can call is a powerful shift.
Excited to see how this fits into scalable image and video generation pipelines.
