Hacker Newsnew | past | comments | ask | show | jobs | submit | silverlight's commentslogin

Google really can’t help themselves but to have some internal re-org kill off a public thing people are actively using. It’s honestly impressive how consistent they are.

The rest of out here watching usage and telemetry to decide where to invest, meanwhile, over at Google…

What if their telemetry shows very low usage? I've seen virtually no discussion of Gemini CLI online.

There are 13,700 forks of its repo on Github.

If anything, I suspect closing the source for their coding agent may have been part of the goal.


If the repo is already forked, what difference does it make if Google supports it or not? The community can just continue development if they so choose

google can kill your account for using an unsupported harness when/if they choose

Complete bs. Gemini models are on OpenRouter just like any other models.

But then you have to pay by the token instead of the subsidized subscription prices

oh, so finally it is about money not open source.

It's the code they develop going forward that matters.

Forks don't mean anything anymore, especially on an AI-related repo. 99% of them are automated.

There's a fair amount of enterprise usage. It's a really good product, despite the Claude hype. Anthropic is a PITA to deal with, and it's slow as shit weekday morning Eastern time.

Gemini CLI for Enterprise will continue.

> If your organization uses Gemini CLI or our IDE extensions via a Gemini Code Assist Standard or Enterprise license, or if your organization uses Gemini Code Assist for GitHub through Google Cloud, your access remains unchanged.


I use all 3 compared to what do you think Gemini CLI is a good product? (my only use case for it now is triple checking specifications for drift inconsistencies beyond that I find it pretty lacking compared to codex and cc).

I think for the average corporate person in a non-software company, the "out of box" experience is better with Gemini. It's also cheap.

Once you get deeper into it, both Codex and Claude have better integrations with skills, etc. I sort of "discovered" skills via GStack and now use a few things, I find Claude's performance infuriating, but it can do more things. I happily pay $200 for Claude now, mostly for my own personal stuff. I think Gemini is better at external data sourcing and coding complex math.

But note this is my anecdotal take, mostly in the context of hobby projects. I'm a journeyman AI slinger at best.


Whats to discuss, it works, it does the job, cant complain.

Except complaints that it's horrible, you mean?

Even thought there are people using, it doesn’t mean they see a future in it. Google is the best when it comes to analytics and trends. If they see a product is expected to fail, which in this case it was, they simply kill it and move on instead of wasting resources saving a sinking ship.

Of course, something could’ve been improved, but that’s just how they operate.

I could be completely wrong though


The usual playbook is they rename it a few times first, then they kill it.

And then later re-use the name for another product which is almost but not entirely unlike the original.

Sometimes it becomes 11 different chat applications among the way.

It's technically a chat application. Except you're chatting with an LLM instead of friends and family.

And building a chat app is how you get promoted at Google.

Another tombstone in Google Graveyard soon [1].

[1] https://killedbygoogle.com/


Thanks

There was a moment about a week ago where Claude went down for about an hour. And right after it came back up it was clear a lot of people had given up and were not using it.

It was probably 3x faster than usual. I got more done in the next hour with it than I do in half a day usually. It was definitely a bit of a glimpse into a potential future of “what if these things weren’t resource constrained and could just fly”.


I had that exact same feeling during the US holidays where I got to enjoy 2x usage limits and everything just seemed to work well


I had terrible results during the holidays -- it wasn't slow but it was clear they were dealing with the load by quantizing in spots because there were entire chunks of days when the results from it were so terrible I gave up and switched to using Gemini or Codex via opencode.


I find that if I have my rabbit's foot and lucky socks on, I win working code ~1.2x more often.


Noticed the exact same thing a few days ago. So much so that I went on twitter and HN to search for “claude speed boost” to see if there was a known new release. Felt like the time I upgraded from a 2400 baud modem to a 14.4 as a kid - everything was just lightning fast (for a brief shining moment).


I would also regret it if they become that fast; right now I can really take a moment to enjoy the hard work the model is doing for me.


https://xkcd.com/303/

the evolution of this xkcd


Seems like a big opportunity for Google to consider keeping Gemini ad-free as a differentiator. They can afford to burn cash on it for a long time to come if they choose to do so.


All enterprise users already pay for it. They’ve included it by force to the base subscription (and about 30% of our company actively uses it, according to in-app stats as an admin).


This is really what's going to kill OpenAI. Gemini bundled in Workspace is pretty good and seems to be making big leaps constantly. Some really smart integration points.


Did the client side JS being infected produce any issues which would have affected end users? As in if a web owner were on an affected version and deployed during the window would the end user of their site have had any negative impact?


No, just the host that was running the package (the exploit was pretty generic and not targeted at PostHog specifically). In fact, so far we think there were 0 production deployments of PostHog because the package was only live for a little bit.


Glad to hear the impact was so muted. Thank you for the response!


Recently used a lot of HaxeUI for a game and it felt similar to this. A lot of components included out of the box that “just work” with some basic styling. And you can use CSS styling on top to further customize as needed. Works across a lot of different platforms as well.


I read that as Haskell and I definitely think HaskellUI would solve all problems. /j


Is this available somewhere to use?


> Seedance 1.0 will be integrated into multiple platforms in June 2025, including Doubao and Jimeng

> https://www.doubao.com/chat/create-video

> https://jimeng.jianying.com/ai-tool/video/generate

https://arxiv.org/pdf/2506.09113


This is obviously going straight to TikTok. The big issue is it's going to open the flood gates on their own platform.

Anyway, if everyone wants to be a content creator, why not charge them for the privilege of that desire? A content creator will forever need AI-generated something. So now we move from "you get to post your content for free" over to "you get to now pay us through this AI-gateway to post your content".


Anyone using one as your daily driver for coding? Or tried and it didn’t keep?


Tried it. It was okay, felt very futuristic. But ultimately there weren’t really any benefits over just using a multi monitor setup that could justify the cost. I thought it’d be nice to have something I could also use for watching movies, but when I watch movies I almost always watch them with my wife, and much prefer that.


Daily driver no, but after the somewhat recent ultrawide monitor update, it is absolutely viable for travel. It's nice not to be constrained by a 14" monitor while on the road, and it's perfectly usable for a couple hours of work (especially if you take the surround seal off and use one of the head strap mounts that take all the pressure off your face).


I tried it, and it works alright. But it's not very comfortable to keep it on while eating/drinking, and you can't wear the thing in public without looking ridiculous.

Personally I found my existing dual-monitor setup to be more ergonomic.


Happy to support this, desperately needs to be changed.


I made one of the original posts on HN about this years ago after hearing about it from my CPA. Both then and now these changes make zero sense to me as a matter of good policy. I am also still surprised at the number of people in tech who either haven’t heard about this or are willfully ignoring it and likely filing their taxes incorrectly.


Somewhat related topic: anyone used something like the Luxedo to do a projection mapping? It seems like it’s more complicated and expensive but I love the idea of doing something cool to the house for Halloween and Christmas without having to lug out a ton of lights and decorations.


There are several open source projection mapper projects that run on a RaspPi, or other platforms, here are a few of them:

https://github.com/arisona/mpm

https://ofxpimapper.com/

https://mapmapteam.github.io/

There are media players and graphics generators that will run on a RaspPi also. Finding a bright enough projector and sheltering it for outdoor use might be the main challenge.

Another approach might be a galvo scanned laser with DMX software to control it, though there may be safety/liability issues with that.


I‘m fascinated by these projection setups ever since I learned they’re a thing! But Jesus is the Luxedo stuff pricey?! I’d love to learn about something similar but in a kinda OSS/DIY kind of style…


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: