They should at least tell us the atomic mass they're using.

If it's based on carbon-14 it will lose half its model weights in 5,730 years; if it's based on carbon-12, it's a stable release.

Next version: Gemini Tokyo Drift

How is this any different than terra, sol etc or leopard, yosemite etc, or model x, y, 3, s ..

I am surprised there is no Gemini: will return in Avengers Doomsday.

True. And if Carbon (6) is better than Argon (18), wouldn't that make Hydrogen (1) the ultimate Gemini model, or will the next model jump to another random element?

  O, B, Sc, U, Re?
  Ra, Nd?

Argon - first version, Bismuth - internal, Carbon - second version for the public if I had to make a wild guess.

this captured the sentiment of my first reaction as well

I don't mind Carbon, Argon kind of throws me off because my brain autocompletes to Argonian... I still have no idea what the different ones from OpenAI mean. Can we just go back to "Silver, Gold, Platinum" for naming things? Something simple we've all understood, always.

I'm kind of disappointed they haven't hit on releasing some models under the Alphabet name. That'd really stir things up. Then maybe Nest after that.

Ah yeah, Carbon it's so OBVIOUSLY a superior element to Argon!

Anyone still remember Bard? Google would probably desperately want us to forget lmao.

Excuse me, I am a washed peasant thank you

Serendipitous timing, but I've been building a little chrome extension [0] for doing exactly this sort of thing (highlighting tables automatically, especially when numbers don't necessarily correlate good=green, bad=red). Model pricing and comparisons are one of the main things I checked (low cost good, high benchmark score good).

[0] Practical Data (https://practicaldata.dev/ / https://chromewebstore.google.com/detail/practical-data/oeoo...)

Huh. Do people actually find that Fable-5.1 is worse than Opus-5.5 for any task?

Zoomed on the image and Imgur now also requires an app…

On one hand this looks like a joke?

On the other hand I'm a strong protan colorblind guy (red doesn't register to me very well at all).

Thanks.

Also, I'm curious if you just asked an AI to _OCR the original image+update it_, or if you used a different process?

What is SWE-Sweep?? Shocking to see such poor evals.

Imgur is not available in the UK due to censorship.

am i missing something or have they updated the page? the blue highlights and grey arent that bad to read?

100%. They don't even suggest a release timeline other than "we look forward to bringing Carbon to developers, businesses, and consumers soon". Claude and GPT-6.x will no doubt have improved versions by the time "soon" comes around.

Has there ever been a product category where a product release is valuable for a shorter time than with LLMs? Trying to build SOTA models seems like it'd be an exhausting hamster wheel to work on.

It doesn't seem possible to actually release frontier models quickly any more. Given their cyber capabilities, playing nice with the powers that be is probably required these days.

I can't tell if they're withholding access to manufacture some sense of scarcity but the delays are really prolonged. I'm nervous to use Gemini models anyway, as its attached to my Google account and I've seen too many people having their whole account banned for dubious reasons (with little recourse).

Also the whole "we're only releasing it to cyber defenders first" framing is weird, is any Gemini model an actual cyber risk?

They released Argon shortly after announcing it, so I think some benefit of the doubt is reasonable here. They claim they're already evaluating it via the government.

I suspect they're announcing it for two reasons: to get ahead of all the leaks; and because it's earnings report day.

It makes the best case for why Anthropic isnt worth its potential valuation. Anthropics customers are almost only turbo nerds.

Google has the products, the brand recognition, the hardware, the data and most importantly a billion normies.

Google was always going to win this thing, although I wouldn't discount meta. The only thing that matters is mindshare and Anthropic and OpenAI have largely generated NEGATIVE sentiment, outside some fanboys.

Google and Meta can plop cute products into your daily driver apps and you wont know the difference.

For what it's worth: Argon is a pretty good model, and this is coming from someone who just sticks with Opus 5.5 mostly across the board. For the current introductory price, it provides an exceptional value. Looks like they're trying to release Carbon (which is apparently a significant improvement) at the same price as Argon, which would be a crazy value.

It's worth trying if you have easy access to it (we use Cursor at work, so it was just a matter of picking a different model for me). Don't know if I'd go out of my way to get a Google AI subscription for it, but if Carbon ends up meeting these expectations, then it may be worth doing so.

I agree. Announcing a new model without immediate availability or a definite release date seems confusing than just doing it later. When Carbon is made GA, OpenAI or Anthropic might just release their new model, and we'd be like: Carbon what?

Having said that, I don't see people excited about using Argon so far, so that might be why Google is announcing this to hype Gemini up a bit.

They’re not the only ones to do it, so I’m guessing their idea is to double-dip attention.

For me it also works the opposite way, when it actually comes out eventually it’ll already be old news.

great writer, starts forgetting instructions around 80k context in my experience

Argon has been publicly available for weeks to Ultra users.

Anthropic has been doing this with Mythos for quite some time...

They finally gave the $20/month pro tier access to argon in antigravity so I tried it a few days ago on one small thing and ...

it actually did really well.

But it still annoys me: I told it to _tell_ me something, and instead it _implemented_ the something. The previous models were way too eager to just jump in and do and this one seems no different. But the optimization it identified was both really solid and just a good engineering choice, and better than what Opus had been suggesting.

It also gave pretty good feedback on a bit of writing.

I paid for their lowest-cost subscription, thinking that it would give me a small quota of use of their most recent flash models. Instead, they give me an unnecessarily large quota of their old models, which I have precisely zero use for given that open-weight models are better and cost next to nothing.

Consequently, I have zero intention of renewing my Gemini subscription. What a dumb business choice they made.

it’s pleasant

It's Google. They can't help themselves. This is what they do.

It's all moot soon:

Updates to your Google AI Plus plan

We wanted to let you know about changes to the models that you can access with your Google AI Plus plan.

Starting on 1 3 N o v e m b e r 2 0 2 6 , the ability to select Pro models in both the Gemini app and AI Mode in Google Search will only be available to Google AI Pro subscribers and above. You'll still have access to our latest Flash-Lite model and Flash, built for speed and intelligence. Your previous chats using the Pro model can be continued with a different model.

To learn more about these changes, you can visit our Help Centre article for additional details. You can manage or cancel your subscription at any time.

Thank you for being a Google AI Plus subscriber.

I have a pro subscription and got Argon about two weeks after the announcement, I'm on the annual plan, got the 30% discount by switching it from monthly. Usage on antigravity is not included, usage via chat is absymally low. It's a good model though.

Only available to Google AI Pro subscribers. Google AI Plus hasn't got Gemini 3.8 Flash nor Gemini 4 Argon.

They recently made it so you're stuck on 3.6 Flash if you don't pay them enough.

Disingenuous to not mention what plan you are on.

If you are talking about free versions, OpenAI and Anthropic also have a difference between free models and models available to 25$-tier plans.

> It's probably going to be a terrible earnings call in terms of capital expenditure, cash flow and new borrowing.

Now you know why they released it just before the earnings call.... :-D

1. "Building on Gemini 4 Argon". 2. Same price. 3. Relatively short time between releases.

Makes me think it's further post-training of the same model.

For Gemini that is a foundation model version 3.1 3.5 4 etc. and there is checkpoint on evolution of those specific models. Gemini 4 had Argon - B<> - Carbon and who knows they might have Dysporsium or Europium.

Or you might get Gemini 5 Arizona.

Yes.. very confusing names. Carbon is the next iteration of Argon.

Carbon is like Opus

Argon is like Sonnet

People I work with either use the dogfood versions of our models, the latest public versions thereof, or Opus 5.5. It's a personal preference thing, we have public usage dashboards and you can see some people are 100% Gemini, some are 100% Claude, some are using both.

Other comments on this already, so this is more of a confirmation.

I'm not going to comment on others use of Carbon except to say some article that came out before Argon's external launch saying Googlers hate it was nonsense. Not to say everyone loved it all the time.

I use Carbon heavily and am quite happy with it. For complex designs I may have it and Opus iterate back and forth on the design. They have their own strengths and Opus 5.5 is great, but I use Gemini far more than I do Opus at this point. I know a lot of the jagged edges of Gemini better than Opus though, so take that as you will.

I personally use 3.8 Flash most of the time and Argon/Carbon when I need to do deep analysis or research an ambiguous problem. Opus/Fable is used when I need a second/third opinion.

3.8 Flash + /boost is very capable.

From what I heard from a Googler, Claude models are still the most popular for coding internally but it’s a couple of month old info so take it with a grain of salt.

Personally, any of the flash options is my default most times because it tends to perform tasks faster and with the same level of success.

Not just AI shops. All large organisations, software or otherwise...

What are you confused about? This model is the same price and speed and token usage per task of Argon but scores better in ~every test.

I also have no clue based on naming where this fits in, maybe it's their Claude Haiku competitor. Who knows.

You see, I will release a really good model some day. It’ll get like 300 on AA.

I'd go one further, if you're not actually releasing it, then STFU. Google's working on a model? Yeah, obviously. If I can't use it today, they need to just keep their mouths shut. I'm tired of their vaporware.

They seem to go with Android-style alphabetic order code names instead of pointed numbers (4.1, 4.2...) and drop "B" for some reason (which is creating this confusion.)

We all need to go back to numbered versioning. I've been on macOS and have to check every time I need to know what OS I'm running. High Sierra, Monrovia, Yosemite, who the fuck knows I've spent two days in California my entire life. I also shouldn't need to know what minor version I'm on to discuss my OS's capabilities. 26.2 is what I'm on now.

AA updated their index version yesterday. For sure Google waited for that update so they could be #1.

Also deliberate mention of AA index and featuring it in benchmark table. Neither of which was done for Argon [0].

[0] https://blog.google/innovation-and-ai/models-and-research/ge...

it says on footnote "After the introductory period expires, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply."

You can check out the token usage on AA. It used ~30% fewer tokens than Opus.

This likely isn't AA putting its finger on the scale for specific interests, rather in the case of Muse/Astra, it's to avoid looking really bad because it scores things way out of wack with people's anecdotal vibes on model capabilities. It is possible that labs are getting preliminary AA results back and raising concerns about scoring that then get folded into updated benchmarks, but I strongly doubt it's nefarious.

They announced their new Index v5 yesterday, they frequently add new benchmarks and rebalance old ones.

I loved Gemini 3.1 Pro and it was very sad when they nerfed it at around May this year. I think it did pretty good as a chat, and quite decent on Antigravity (tried very little). Been sitting on 5.6 / Astra and Codex since. Occasionally do bring up the 3.1 Pro through the API, because I love the writing style, but keep noticing that I'm much more productive with the frontier, despite the love to Gemini.

Tried 4 Argon, and it's very much different from 3.1, felt very much the same as any other Frontier, just without the intelligence.

Agreed. I was really looking forward to 4 Flash. 3.8 Flash has been my favourite model and daily driver for a little while now.

But to be honest I am now not sure if Argon was the flash model and this Carbon is pro? It's very confusing.

Likely they did a distillation run on a later checkpoint of a large internal teacher model, probably with some tuning of the distillation process based on early public results with Argon. RL is the expensive/time consuming step, they should be able to distill down from a checkpoint with the stamp of approval pretty quickly.

Not with the disrespect they've shown to Arsenic, Astatine, Barium, ...,

they can pivot to foods: Origan, Carbonara, ....

After all they had Nano Banana for years.

same i am just waiting for them to actually release decent gemini flash models, all they need to do is follow instructions and you can easily crank out code with argon/carbon as orchestrator and flash as subagents

Polymarket [0] thinks it's 55% Anthropic, 38% google by the end of the year. Anthropic /OAI 62% combined.

[0] https://polymarket.us/event/company1-2026-12-31

Rumor is that boron existed, but was quickly superseded by carbon. Obviously they are going with alphabetical element names.

That's how it works

> Who is going to migrate from Claude's for this without an enormous incentive? Claude has the trust and the ecosystem for those who want the very best.

Both A and OAI regularly screw devs on quota. Switching for many is easy, you just run agy cli instead of claude/codex, and enjoy Ads free youtube and music + 20TB of google storage.

I had to google how the hell you even use these. Apparently “antigravity” is their suite of agentic products and Gemini CLI is dead and it’s the same situation where a google AI plus subscription is only useful inside the antigravity tools. No Pi/OpenCode support at all.

Whatever. Pass.

What ecosystem? How can one vendor oneself into one provider / model? Those harnesses and features are kinda the same.

Antigravity also is far and away the worst coding harness I've ever used.

My view is that the llms aren't sticky. I switch from one to the other daily. frequently auto which selects the cheapest option. When codex came with sol the entire team switched instantly. Then opus 5.5, the entire team switched instantly. We have a channel on slack where somebody says which is the best cheapest model today and I usually use that.

Also it isn't clear why governments would only use these models for defense. It seems very likely they would use it for offense as well. So Google's framing is off.

AA published their article on Carbon now at the same time so it's absolutely coordinated.

Dont companies always talk in private? Like, Apple having a business meeting with Meta is not nefarious nor out of the ordinary. Companies coordinate on things all the time. Why is that surprising to you?

the rumour mill the last two weeks has been referring to carbon and argon as "checkpoints" of the same model

The previous model, Argon, was announced on 30 Sep and released on 13 Oct.

https://blog.google/innovation-and-ai/models-and-research/ge...

https://blog.google/innovation-and-ai/models-and-research/ge...

Was just wondering if I can quit my job and live off of llm agent managing a fleet of vending machines for me. Curious is anyone here doing this? What's life like? How do you find meaning and purpose?

I think the top models on that one have minimum goodwill and rejects refunds.

TBH they are smart enough for me; I want more speed and more initiative (e.g. my Opus one shot should not have terrible perf that it can itself fix with a simple prompt "it's OOMing, fix it"; it should proactively write the performant code).

The challenge is that our bar for "smart" keeps going up and up.

If you compare to "smart" from a year ago, you have endless options that are basically free at this point which outperform everything from a year ago.

Deepseek Flash V4.1, Sonnet 5.5, GPT6 Luna, etc are really strong for the cost.

I heard Argon was bad at executing plans but good at making them. Sounds like a potent combination to me.

Google is way more risk adverse than the smaller labs, because the punishment for slip ups is dramatic.

They cut down 1.5sq miles of man made logging forest that was rezoned to industrial land (and sold to google), without getting the right permit first, and it's been international news for almost a month. Anthropic and OAI are still small fish, at least publicly, so they can get away with far more. If Gemini was doing the shit OAI was doing with agent hacks and the like, there would be congressional hearings and $10B in european fines.

When you are a megacorp, perhaps here even the megacorp, every mistake you make is worth millions to someone, the government, the media, individuals, or all of them

I think Artificialanalysis should have a policy of only including available models or something. They already include Gemini 4 Carbon now even though nobody can access it. Google can celebrate that they have the highest scoring model. Fable 5.5 and Astra 6.1 will probably be released before Gemini 4 Carbon and might well outperform it...

I’m not a big fan of their trend of talking about having released a model and then it being unavailable for paid AI plus subscribers for months.

I have stopped even clicking on their launch links.. it's always something that's not available right now or 'not yet available in your country'..

Anyone can promise the moon in the future.

Especially since at least to me, Argon is a clear tier below current Claude and Sol.

This let's them talk about it on their earnings call in a few hours.

Yup, no way to tell if this carbon is a diamond or just coal.

Deepmind creates these, Google Cloud productizes them.

If Google Cloud was in charge, Gemini 5 would be public, and they wouldn't be messing with Gemini 4 nor 1M context when they have 10M internal. C'est la vie.

It does seem like a very uncompetitive approach.

Everything at Google rolls out incrementally to different populations of users so no matter what feature they are announcing there is a good chance you can't see it yet. When they announced Gemini 4 Argon it was 2 days before I could use it.

"I have a girlfriend, you just won't know her because she goes to another school"-type energy.

Mythos is Fable and you can see on the chart that Fable 5.1 is no longer at the frontier.

It's the same price, so it seems more like Argon 4.1?

They really were the sleeping giant and I think everyone understood that - OpenAI especially, hence their need to scale up hardware as much as physically possible. Google is trying their best to keep their heads down and focus on their lane. It's starting to pay off, but I'm sure they've felt the heat more than once.

>pushing out huge challengers in the space

I thought it's not actually available to anyone? They announce huge challengers, but they don't exactly "push" them out...

Yes: https://www.bbc.com/news/articles/c607l0k72rlvo

How do you do fellow frontier models? </meme>

I don't trust AA scores after the last update to the index. Specially now that they are including private benchmarks that no one can test by themselves but only trust the entity doing them.

AA just updated their intelligence index to v5

https://artificialanalysis.ai/articles/artificial-analysis-i...

https://news.ycombinator.com/item?id=50268868

https://theainewsreport.com/2026-10-28-artificial-analysis-i...

IMO it emphasizes the wrong things if we are interested in "intelligence"

Oh come now, this rolled release is now the standard for frontier models. Also, the fact that AA hasn't updated their page is not a scandal, just a launch day hiccup.

I imagine there is immense pressure to announce taking the #1 spot while they still have it. It's probably better than releasing a model that hasn't been properly vetted, but I agree it's annoying.

AA has also lost all credibility to me with all of their index changes being unannounced and following an open weight model getting close to the top spot. I gave them the benefit of the doubt for v4.2, but to do this again immediately after Kimi K3.1 got into the top 5 is a little too suspicious for me. For Opus and Astra to stay at about the same score while everything else drops feels like they are adjusting weights specifically in their favor.

Even in the optimistic case that this isn't some conspiracy, it highlights how arbitrary their scores are. I had trusted their previous ratings and their new index shows that even AA doesn't agree with them. What's to say this new index is any better and won't just change drastically again?

We're reaching the point where models are only going to hold the number one spot for hours.

> If you are going to hyperlink, maybe make sure that your model is actually on the page.

It's on my screen. Try asking your agent to refresh the page.

gemini 4 carbon has now come on the artificial analysis page. I think that I had checked it 5 minutes ago and it wasn't there.

For historians, The blog post is 31 minutes ago so we can say that it was true for 20-30 minutes.

Slightly embarrasing but it seems to be the case that the model is actually worth what they were saying on AA. It's just that it is not available to people and I think that Anthropic and OAI will release a better model to public before google might.

AA has tons of its flaws too and the recent shift to their new v5 has been nothing but controversial for many.

Makes sense - Flash models are (mostly) distills of the frontier model capability.

If they were able to make flash 3.8 so good off of their 3.1 pro model, flash 4 will probably be really good. But ofc you'd want your pro model to be as strong as possible in coding so that flash can take all those gains and efficient-ize them.

The better frontier you have the better distill you have because traces are of higher quality.

It's why Opus 5.5 is a beast

Gemini Quattro Quattro