Interesting but federal tax credits for heat pump water heaters is ending next month [1]. It looks nice though.
For people wondering what's up with the "150 gal virtual" capacity, it's actually a 47 gal water heater but it heats water to a really high temperature (if needed) and mixes it with a thermostatic valve to output the desired temperature. For example if cold water was 15 degrees C and you want the hot water to be 40 degrees C, it can triple its "capacity" by heating it up to 90 degrees C and then mixing one part hot water with 2 parts cold water to give you three times as much hot water. Of course, this comes with a grave penalty to efficiency so you should only do it for "party mode".
I had one of these types of water heaters in an old apartment. Thing worked kinda okay-ish? The main issue was that it was electric and if you used the hot water for really any time at all (bath, dishwasher) then it would pop the fuse from current draws and overheating. Likely this was some issue with the water heater and the electrical system of the apartment.
Still, really annoying when half way through your bath, the fuse would pop, colder water now coming out, and you'd have to completely dry yourself to reset it and wait for whatever got too hot to cool down enough to get the other half of the bath going again.
They must be doing something different to get that much capacity. There’s a bunch of water heaters with mixing valves but they don’t talk about that much capacity. What worries me is, doesn’t higher temperature make your tank rust faster?
I was surprised at how cold "hot" water actually was. I thought it was 60-70, but apparently what feels "hot" is around 45-50. Especially for me, that finds anything beyond my shower's "middle" heat as uncomfortably hot, I must be showering with around 40 C water, which is basically "hot day" hot.
Going less than 55 °C tap temperature / 60 °C in the tank is bad though, at least in larger installations - otherwise, you risk legionella and other microbial infestations [1].
It's supposed to be HOT water, not lukewarm... I have my water heater set to 150F because it makes the hot water last longer, especially in the winter when the incoming supply is barely above freezing, but that doesn't make the tank bigger.
LLMs can't "look" at the rendered HTML output to see if what they generated makes sense or not. But there ought to be a way to do that right? To let the model iterate until what it generates looks right.
Currently, at work, I'm using Cursor for something that has an OpenGL visualization program. It's incredibly frustrating trying to describe bugs to the AI because it is completely blind. Like I just wanna tell it "there's no line connecting these two points but there ought to be one!" or "your polygon is obviously malformed as it is missing a bunch of points and intersects itself" but it's impossible. I end up having to make the AI add debug prints to, say, print out the position of each vertex, in order to convince it that it has a bug. Very high friction and annoying!!!
Cursor has this with their "browser" function for web dev, quite useful
You can also give it a mcp setup that it can send a screenshot to the conversation, though unsure if anyone made an easy enough "take screenshot of a specific window id" kind of mcp, so may need to be built first
I guess you could also ask it to build that mcp for you...
I had some success providing screenshots to Cursor directly. It worked well for web UIs as well as generated graphs in Python. It makes them a bit less blind, though I feel more iterations are required.
Kinda - Hand waiving over the question of if an LLM can really "look" but you can connect Cursor to a Puppeteer MCP server which will allow it to iterate with "eyes" by using Puppeteer to screenshot it's own output. Still has issues, but it does solve really silly mistakes often simply by having this MCP available.
You can absolutely do this. In fact, with Claude Anthropic encourages you to send it screenshots. It works very well if you aren't expecting pixel-perfection.
YMMV with other models but Sonnet 4.5 is good with things like this - writing the code, "seeing" the output and then iterating on it.
Claude totally can, same with ChatGPT. Upload a picture to either one of them via the app and tell it there's no line where there should be. There’s some plumbing involved to get it to work in Claude code or codex, but yes, computers can "see". If you have lm-server, there's tons of non-text models you can point your code at.
There has got to be some firmware bugs causing high power usage, but also the CIX CPU is fabbed on TSMC 6 nm process which is a couple generations old.
Imagine if someone clicks the deploy button when you're in the middle of typing something and then the service goes down due to a syntax error. To prevent this, we will need some sort of way to set a global lock to indicate that "I'm not done typing yet" and you can only deploy once everyone has released this lock.
Or you don't deploy unless it makes it through at least testing, and a build started while someone was editing the code would probably fail fast unless you coincidentally hit the button right when it's valid, but wrong, code.
You can likely install Windows on Steam Machine if you so wish, and then it would actually be a fairly competent mini PC while having great and silent cooling. However, I suppose most casual gamers aren't savvy enough to tinker and install their own OS.
Having a single big fan cool a massive heatsink (that is hopefully very quiet) can legitimately a good reason to get this over building a typical SFF PC, which often runs hot and loud. It sorta reminds me of the trashcan Mac Pro. I myself have a sandwich style case with an RTX 5070 in it which is quite loud under load.
> if they have anything figured out besides "collect spatial data" like imagenet
I mean she launched her whole career with imagenet so you can hardly blame her for thinking that way. But on the other hand, there's something bitter lesson-pilled about letting a model "figure out" spatial relationships just by looking at tons of data. And tbh the recent progress [1] of worldlabs.ai (Dr Fei Fei Li's startup) looks quite promising for a model that understands stuff including reflections and stuff.
> looks quite promising for a model that understands stuff including reflections and stuff.
I got the opposite impression when trying their demo...[0]. Even in their examples some of these issues exist like how objects stay a constant size despite moving. Like missing the parallax or depth information. Not to mention that they show it walking on water lol
As for reflections, I don't get that impression either. They seem extremely brittle to movement.
For people wondering what's up with the "150 gal virtual" capacity, it's actually a 47 gal water heater but it heats water to a really high temperature (if needed) and mixes it with a thermostatic valve to output the desired temperature. For example if cold water was 15 degrees C and you want the hot water to be 40 degrees C, it can triple its "capacity" by heating it up to 90 degrees C and then mixing one part hot water with 2 parts cold water to give you three times as much hot water. Of course, this comes with a grave penalty to efficiency so you should only do it for "party mode".
[1] https://www.energystar.gov/about/federal-tax-credits
reply