Opinions are my own.
I'm not sure this list is correct. Number 4 is especially wrong, since Skills are available with the launch of Antigravity 2:
https://antigravity.google/blog/introducing-google-antigravi...
2,470 karma · joined January 16, 2012
I'm not sure this list is correct. Number 4 is especially wrong, since Skills are available with the launch of Antigravity 2:
https://antigravity.google/blog/introducing-google-antigravi...
In 2026, a single minimum wage income can barely survive by themselves with no saving in most part of the US.
> Antigravity was the only autonomous agent that implemented the Pantheon’s signature interior ceiling pattern: repeated square coffers visible through the oculus.
The article also includes a video showing the patterns visible through the roof oculus.
I think the biggest winner of this might be Google. Virtually all the frontier AI labs use TPU. The only one that doesn't use TPU is OpenAI due to the exclusive deal with Microsoft. Given the newly launched Gen 8 TPU this month, it's likely OpenAI will contemplate using TPU too.
For agentic work, both Gemini 3.1 and Opus 4.6 passed the bar for me. I do prefer Opus because my SIs are tuned for that, and I don't want to rewrite them.
But ChatGPT models don't pass the bar. It seems to be trained to be conversational and role-playing. It "acts" like an agent, but it fails to keep the context to really complete the task. It's a bit tiring to always have to double check its work / results.
Short form pop content like TikTok doesn't give my brain enough time to engage the thinking muscle.
I think it's better to identify the characteristics of the media we consume, rather than lumping all of them together.
https://en.wikipedia.org/wiki/Tone_(linguistics)
The modern Vietnamese alphabet was developed in 17th century (so it's not a transliteration) with tonal marks as a core feature. The writing language is very phonetic. Within a region with similar accent, if you hear a word, you can write it. And if you see a word, you can pronounce it.
The tonal marks are very important to the language. It allows for rich poetic rules that makes Vietnamese poem fun and musical to read:
And those text got into the training set: https://www.reddit.com/r/MandelaEffect/comments/qbvbrm/anyon...
- The USD is definitely losing value. That also means stocks from US companies would be cheaper from a foreigner's point of view.
- That means it represents good investment opportunity as long as the fundamentals of those companies are not affected too much (e.g. AI companies not directly affected by workers' raid, or pay tarrifs). Nothing is contradictory here.
The printer is never "done" :). But there are plenty check points where it's "pretty good".
For example, here is my rough timeline:
- I sourced the parts and built it. Took around 4 weekends.
- The initial tuning took a while (like a month). But this was very fun. I tried almost all the Slicers. I fixed constructions issues (square angle, deracking, belt tuning, ...). After this step, the machine becomes "good enough". I can print various parts in the house and I was satisfied with the quality.
- I started pushing for speed and redid many parts of the printer. I learned about various limitations (like Flow Rate is the real limit for speed). This phase last a long time for me (like a year). I ended up replacing like 75% of the printed parts with CNC parts. During this time, the printer is still online and printable.
- I didn't modded the printer much after that. I found my sweet spot between speed / quality. I want to mod it with a 120W Hot End heater to increase the Flow Rate (already bought it), but it's not quite a necessary thing. It's more for fun. The tinkering goes on as long as you feel it's fun. But I wouldn't say you _need_ to tinker to _keep_ it working.
> Is there a point after which it “just works”? Or is it always going to be more like “it’s great! I just need to tweak the blah blah setting every time and retighten the frobnitz every 3 prints, no big deal really!”
After the first tuning phase, the Voron was "just works" for me. Or at least, if there was any issue, I could immediately tell what went wrong. And no retightening needed so far except one time the printed feet cracked (that was the reason I switched to CNC aluminum parts).
Edit: I built a large Voron (350mm), and it is really _heavy_ (almost full metal in my case). That's why the printed feet cracked. Beside that, maintenance is almost zero. I don't even wash the spring steel bed. Just click print and walk away.
I built myself a Voron, and it's an amazing learning experience. I learn how things work, and the trade offs. I get to pick and replace the exact parts I want. I design my functional parts knowing exactly the printer's capability. There is something very fascinating about it. You can look at a print, and can tell different issues at a glance because you have seen (and fixed) them while you built and tuned the printer. The majority of 3D Printing quality issue are due to Hardware constructions / trade offs, and not Software (slicer settings..). Without building a printer from scratch, it's hard to tell the root cause.
https://vorondesign.com/voron2.4
- Fully open sourced
- Repairability and updatability. Lots of fun mods.
- No phone home / privacy issue like Bambu
I think before going down the rabbit hole, it's best to make sure you have a clear answer for this question: Do you care about the learning / tinkering / optimizing part, or do you care more about "it just works" printing?
- Many recommendations in this thread is for the "it just works" printing case. The top candidates are Bambu, Creality, and Eiegoo. The quality is good for most cases.
- If you're an engineer and into tinkering like me, you would be much happier with a Voron v2. Depending on your effort, you can match Bambu's quality, or _greatly_ exceed it.
Regarding Slicer, don't worry much about it. You can learn one very fast. The top ones are Cura and Orca Slicer. I use them both, and they have pros / cons. Personally on my Voron, a well tuned Cura profile yield better result. But Cura is missing one important feature: it can't limit the speed based on Flow Rate.
Another quick tip:
- Take the advertised number with a grant of salt. For example, many printers advertised 600 mm/s print speed. The mechanical frame may be able to handle 600 mm/s, but the Hot End is the limit of the build (e.g. it can't melt material fast enough, friction, the ability of extruder motor to quickly change speed, etc).
Hope you have a great time!
- If Google is confident the source is spam (e.g. known spam center). The call is blocked outright. It still has a log that a call from this has been blocked.
- If it only suspects spam, Google will answer the call using AI bot, something like "Hi, I'm Google Assistant on behalf of XYZ, what's the call for?". The phone shows that it's screening a phone call, but doesn't ring. Only after the caller gives the reason, and it passes the spam check, then it rings the phone. You can always pick up the call early if you recognize what they talks about (from the transcript)
- If it's known good source (contact list, doctors,...), then it rings directly.
So far, the rate of spam I got is 0, and it screens about 20 calls a month.
I'm using the Pixel Fold at the moment, and it's the best phone I've used to date. It's something I didn't know I want until I have it.
Quick review:
- The phone construction feels good on hand and in the pocket. The screen is beautiful.
- When folded, functionality-wise it's like previous Pixel (beside looking better with the metal edge). I spent about 75% of my phone time in this mode. Also no notch!
- When unfolded, you have access to much more screen real estate. I didn't realize how this dramatically improve reading documents / browsing the web. Things that were unusable (like opening Google Sheet) is now much more comfortable. You can also do split screen, where you keep 2 apps on at the same time (todo list + message)
- The weight feels solid. The fold mechanism is solid. Battery ~50% per day with no battery saving. Camera is good as usual.
Software:
- I've mentioned before on HN, the Spam Screening feature singlehanded keeps me in the Pixel ecosystem. No spam call at all.
- Android Auto is solid
- Gemini is a gentle surprise, especially with how it's easy to interact with the "current phone screen".
Review caveats:
- I don't game on the phone or any CPU intensive tasks. It's plenty fast for me so far.
- I don't use the speaker (only use bluetooth headphones)
Attack surface aside, it's possible that this AI thing might cancel a meeting with my CEO just so it can make time to schedule a social chat. At the moment, the benefits seem small, and the cost of a fallout is high.
On the other hand, both `ruff` and `ty` are about code style. They both edit the code, either to format or fix typing / lint issues. They are good candidates to be merged.
It seems that you may need the "Ultra" version if you want strict prompt adherence.
It's an interesting strategy. Personally, I notice that most of the times I actually don't need strict prompt adherence for image generation. If it looks nice, I'll accept it. If it doesn't, I'll click generate again. For creativity task, following the prompt too strictly might not be the outcome the users want.
- For FrontEnd or easy code, it's a speed up. I think it's more like 2x instead of 3x.
- For my backend (hard trading algo), it has like 90% failure rate so far. There is just so much for it to reason through (balance sheet, lots, wash, etc). All agents I have tried, even on Max mode, couldn't reason through all the cases correctly. They end up thrashing back and forth. Gemini most of the time will go into the "depressed" mode on the code base.
One thing I notice is that the Max mode on Cursor is not worth it for my particular use case. The problem is either easy (frontend), which means any agent can solve it, or it's hard, and Max mode can't solve it. I tend to pick the fast model over strong model.
I understand it's a sale tactics. But it seems not forthcoming, and it's hard for me to trust the rest of the claims.
I was excited, then I read this:
> Send up to 1,000 messages per day—enough for 3–4 hours of uninterrupted vibe coding.
I don't mind paying for services I use. But it's hard to take this seriously when the first paragraph claim is contradicting the fine prints.