News for hackers. Stuff that matters.
Re:Why is Gemini 4 Carbon taking so long? by tristanj on Wednesday November 18, 2026 @12:40AM (#50542293)
It's pretty obvious that Gemini 4 Carbon is not token efficient. It also has relatively low tokens-per-second compared to other models. I can think of several reasons why this is the case:

1. Google has huge amounts of cash and compute. Their goal was to train a model that that has the highest quality & frontier performance, not create a model that's efficient to run and serve. They have other models for that, like Hydrogen.

2. Google created Carbon without caring for inference costs. DeepMind is subsidized by the rest of Google's empire. Selling inference is an insignificant part of Google's revenue. But for Anthropic and OpenAI, selling inference is their entire business model. Anthropic and OpenAI have enormous incentive to optimize model serving and inference costs, far more so than Google does.

Re:Why is Gemini 4 Carbon taking so long? by fsh on Wednesday November 18, 2026 @01:28AM (#50542574)
4. Google did a worse job than Anthropic and OpenAI.
Re:Why is Gemini 4 Carbon taking so long? by dpe82 on Wednesday November 18, 2026 @01:36AM (#50542626)
Well sure, but parent is trying to reason about why.
Re:Why is Gemini 4 Carbon taking so long? by andy_ppp on Wednesday November 18, 2026 @02:36AM (#50542998)
It’s unlikely to be deliberate tbh, it’s more likely the data, algorithms and the training environments aren’t quite as good as Anthropic, Open AI is a lot closer to Google than people want to admit.
Re:Why is Gemini 4 Carbon taking so long? by exe34 on Wednesday November 18, 2026 @05:15AM (#50544358)
Why would data and training environment lead to frontier intelligence but slow?
Re:Why is Gemini 4 Carbon taking so long? by andy_ppp on Wednesday November 18, 2026 @05:48AM (#50544619)
Because the environment did not reward being as efficient as other models?
Re:Why is Gemini 4 Carbon taking so long? by exe34 on Wednesday November 18, 2026 @05:53AM (#50544671)
Oh so it's using more tokens to do the same work. Make sense, thanks!
Re:Why is Gemini 4 Carbon taking so long? by LoganDark on Wednesday November 18, 2026 @05:33AM (#50544504)
Why would a search engine that's had nearly the entire internet for decades not have the data of a lab that's only been scraping for a few years?
Re:Why is Gemini 4 Carbon taking so long? by andy_ppp on Wednesday November 18, 2026 @05:49AM (#50544625)
Because Google are probably more hamstrung by bureaucracy and little things like the law.
Re:Why is Gemini 4 Carbon taking so long? by LoganDark on Wednesday November 18, 2026 @06:31AM (#50545000)
Since when? Google recently disobeyed some legal requirements because "haha, couldn't catch me / too late now bitch". (This is when they cut down a forest for a datacenter)
Re:Why is Gemini 4 Carbon taking so long? by tibbar on Wednesday November 18, 2026 @07:00AM (#50545234)
Internet data is only good for some parts of model training. That is why there is now a huge cottage industry that creates custom datasets, RL envs, etc.

Incidentally, Google is often considered to be extremely good at pretraining, which is just the thing internet data is good for.

Re:Why is Gemini 4 Carbon taking so long? by andy_ppp on Wednesday November 18, 2026 @08:29AM (#50546085)
Yes I'm very much talking about data used in the RL section, I'm sure Google are indeed very good at pre-training.
Re:Why is Gemini 4 Carbon taking so long? by PaulHoule on Wednesday November 18, 2026 @08:12AM (#50545900)
The web is mostly trash. Go look at commoncrawl!
Re:Why is Gemini 4 Carbon taking so long? by teknologist on Wednesday November 18, 2026 @01:29AM (#50542577)
Also, Google uses their own TPU architecture rather than GPU/Nvidia.
Re:Why is Gemini 4 Carbon taking so long? by zarzavat on Wednesday November 18, 2026 @01:41AM (#50542663)
It's 1. Carbon was announced the day of Google's earnings call. It's a PR stunt.
Re:Why is Gemini 4 Carbon taking so long? by kyrra on Wednesday November 18, 2026 @08:33AM (#50546126)
Googlers had been testing carbon for some time prior to the announcement.
Re:Why is Gemini 4 Carbon taking so long? by aberoham on Wednesday November 18, 2026 @03:04AM (#50543221)
Google's frontend web servers routinely time out connections from agy (the antigravity cli) when using Carbon, wasting tons of work. Like any effort level more than "low" is completely unusable.

It looks as if Carbon often takes more than 240 seconds thinking, which I'd guess appears as a silent connection to Google's load balancers, which then drop what it thinks is a stalled connection.

One hopes that Gemini's serving stack is smart enough to cancel these torn-down connections, but maybe they're not and this is leading to all sorts of ghost TPU load.

Disclaimer: We spend millions on GCP per month and have a dedicated customer engineer. Even they are horrified at the glacial pace of the Gemini fixes.

Re:Why is Gemini 4 Carbon taking so long? by p_l on Wednesday November 18, 2026 @03:24AM (#50543408)
This has been acknowledged by recent "agy" release notes, it was timing out in client not in backend
Re:Why is Gemini 4 Carbon taking so long? by kingleopold on Wednesday November 18, 2026 @05:09AM (#50544307)
this happened even before agy btw, their CLI was subpar level for months even before 3.1 pro release and now too
Re:Why is Gemini 4 Carbon taking so long? by yard2010 on Wednesday November 18, 2026 @06:23AM (#50544944)
I never got any google model to work using their infra - 429, billing errors and timeouts. As someone wrote here a few months ago - it's an exercise in managing frustration. So I had to resort to openrouter.
Re:Why is Gemini 4 Carbon taking so long? by embedding-shape on Wednesday November 18, 2026 @06:44AM (#50545116)
Lots of Google tooling is the same for me, simply unusable whenever I try even simple stuff.

I think no one tests this as a "non-Google employee" or something, they all have specific testing backends or something, and once deployed to the public, there is no one responsible for the experience of non-Googlers and no one testing it, that's why most of the Google stuff has such poor UX because of hundreds of small cuts all over the experience.

Re:Why is Gemini 4 Carbon taking so long? by antihipocrat on Wednesday November 18, 2026 @07:49AM (#50545686)
Google never had the mandate to align design standards across internal and external tooling. BigQuery and some of the other analytics tools are great though, I'd be surprised if staff didn't use the same internally - maybe someone here can confirm?
Re:Why is Gemini 4 Carbon taking so long? by kyrra on Wednesday November 18, 2026 @08:32AM (#50546116)
What specific things are like this? If you have a complaint and it's something I can reproduce, I can attempt to file a bug for it.
Re:Why is Gemini 4 Carbon taking so long? by vintermann on Wednesday November 18, 2026 @03:33AM (#50543497)
So in short, as is usually claimed about Google products from bazel to tensorflow to Kubernetes: they may work better inside Google than outside of it.
Re:Why is Gemini 4 Carbon taking so long? by npn on Wednesday November 18, 2026 @07:13AM (#50545357)
4.they are so behind that they can not create effective datasets like other frontier labs do. So instead they do all the tricks the Chinese labs do: using RL with the single purpose of benchmaxxing the AA score.

Look how all Chinese models, or haiku/luna class US models, you can see the same pattern: they think too much and only work well with a subset of tasks

Re:Why is Gemini 4 Carbon taking so long? by Doohickey-d on Wednesday November 18, 2026 @07:25AM (#50545477)
Which is weird, because Google is in the best position for getting training data: they have YouTube (which I guess is why Google's models are good at multimedia), as well as the Google crawler. Which effectively websites cannot opt out of if they want to be Google searchable. So Google's bot probably is the _only_ bot (scraper) that hasn't been blocked from a large portion of the internet.
Re:Why is Gemini 4 Carbon taking so long? by giancarlostoro on Wednesday November 18, 2026 @07:51AM (#50545716)
Not only that, they pioneered some of the key technologies... Who isn't still on transformer models? Also funny when you consider Meta is in a weird similar boat, introduced key things, so far behind the frontier.
Re:Why is Gemini 4 Carbon taking so long? by aswegs8 on Wednesday November 18, 2026 @08:36AM (#50546153)
The innovator's dilemma. Large companies are structurally worse at innovation.
Re:Why is Gemini 4 Carbon taking so long? by adventured on Wednesday November 18, 2026 @09:19AM (#50546637)
To go along with that they're extremely slow (massive orgs) and don't feel their existence is at stake.

If GPT slips meaningfully behind, OpenAI is dead. The same goes for Anthropic. Both would die extremely quickly.

If you're within Google that is not the sense you have about the current context, nor within Meta or ByteDance or Amazon or Microsoft. They certainly view all of this as very important, they don't think they'll die within a year or two if they don't stay on top of the frontier. Meta has run from behind endlessly, as has Google, and their businesses are doing just fine (for now). If OpenAI were in that position, there would be no reason for them to exist and they would be dead by now.

Microsoft likely doubts their existence is at risk just because they can't field a frontier model. For now they don't see Office / Windows being killed off as a short-term risk (plus they have Azure and other products).

AWS is extraordinarily difficult to replicate, it's more akin to trying to compete with dozens of steel mills (industrial infrastructure) than a mere piece of software. And the rest of Amazon's cash spigot is advertising on its retail platform, which is facing close to zero threat from the frontier race, so Amazon can take its time and endlessly screw-up on AI (and so it has).

A frontier LLM can't manufacture a smartphone or laptop for you, and certainly can't manufacture 100 million of them with a zillion dollar manufacturing contract with TSMC et al., Apple isn't particularly worried. They want intelligence on their platform, they're not overly concerned about needing to own a frontier model. They can afford to be patient and get it right.

Capital is more powerful than AI. The capital owns the GPUs the AI requires, the capital owns the datacenters, the capital owns the grid, the capital owns the generation. Apple, Microsoft, Google, Amazon, Nvidia control the capital, financially they're far more powerful than most mega banks these days. And if Apple needed $100 billion for a project, it's one human phone call to Berkshire Hathaway. For now AI is feeble standing next to capital.

Re:Why is Gemini 4 Carbon taking so long? by pfannkuchen on Wednesday November 18, 2026 @09:48AM (#50546975)
Well “they” didn’t, people who work for them did. Are any of those people still at Google?
Re:Why is Gemini 4 Carbon taking so long? by efromvt on Wednesday November 18, 2026 @07:57AM (#50545776)
I think raw base text training data may be decreasingly important vs good RL environments/gyms, and maybe less clear that Google wins by default there?
Re:Why is Gemini 4 Carbon taking so long? by htrp on Wednesday November 18, 2026 @09:47AM (#50546968)
Google legal refuses to allow them to do so.
Re:Why is Gemini 4 Carbon taking so long? by owkman on Wednesday November 18, 2026 @08:52AM (#50546324)
Seeing how every new benchmark that has come out has Carbon doing well, I'd have expected the benchmaxxing allegations to die out. But they've been ok incredibly persistent. It's jagged like all frontier models. You could argue about it's warts, but benchmaxxing is unfounded. Did it benchmaxx every public and private dataset? Is that what RL is perhaps?
Re:Why is Gemini 4 Carbon taking so long? by moffkalast on Wednesday November 18, 2026 @07:43AM (#50545621)
Maybe they should be more efficient and sell fewer ads.
Re:Why is Gemini 4 Carbon taking so long? by Danywalery on Wednesday November 18, 2026 @07:48AM (#50545674)
Google has a huge flow of user data; they were the first to figure out about the Transformer, yet they’re lagging behind in the race. This is very strange, although their flash models are quite good, they’re expensive compared to the Chinese ones.
Re:Why is Gemini 4 Carbon taking so long? by kyrra on Wednesday November 18, 2026 @08:55AM (#50546358)
1) "Gemini 4 Hydrogen is adapted from Gemini 4 Carbon." From https://deepmind.google/models/model-cards/gemini-4-hydrogen... . Also, the Agy team is pushing is that Carbon is a great planner, and then use hydrogen as the worker. They want people to use it.

2) lots of people inside Google use carbon. If they didn't care about it, it wouldn't have been pushed on all of Google.

3) Deepmind is competing with cloud and others. they are the only Frontier Lab that actually has to compete with other business units within an org for compute. GCP and others want that compute as well.

(Googler, opinions are my own)

Re:Why is Gemini 4 Carbon taking so long? by zamadatix on Wednesday November 18, 2026 @09:47AM (#50546963)
1. Small, efficient models are often derived from "the best" models. That doesn't mean "the best models" are efficient too.

2. GP said they don't care about cost, not they don't care about the model

3. Competing for existing compute + being able to assign the projects own capex investment for new compute provides more compute than needing to assign all compute via new capex investment. But it's hard to say if that's actually the type of case or not unless Google wants to tell everyone publicly - so I'd take that point with a grain of salt in either way.

Re:Why is Gemini 4 Carbon taking so long? by serioussecurity on Wednesday November 18, 2026 @09:50AM (#50546993)
(2) isn't right. Cloud is on track to become the largest source of revenue and a lot of that comes from selling inference.