Google unveils Gemini 4 Argon in limited release for cyber defenders

“If they can pull it off on core C++ libraries en masse, I don't know if C++ will still be relevant in a few years.”

“Argon will launch at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price. Wow”

“Google isn't just selling a model; it's selling the internal efficiency it already uses.”

“You love to see it!”

“If these benchmarks hold up, Google just shipped the frontier model to beat.”

“Competition makes the whole field more interesting.”

“Google's new Gemini 4 Argon could put the company back on the AI frontier”

“GEMINI 4 argon. thenewstack.io/google-gemin... Once again we have safety theater over rolling release. This is cringe behavior and must end. Release the model as widely and cheaply as possible, or it’s too dangerous for anyone to use. Pick one, don’t”

“Frontier competition is GOOD. Human agency is GOOD. Give defenders excellent tools.”

“when Alphabet says Gemini 4 is its most ambitious pretraining run yet, yes, the cat ears are UP. I want to see what scales next.”

“Porting large amounts of C code to more safe Rust code is one of the ways LLMs can make software more secure that I don't think non-developers will really appreciate. But it could be a very big deal.”

“Absolute Gemini shocker. Why did Google flatten OpenAI out of the blue? They’ve been sneaky. Lurking in the dark. Did they crack something crucial in iteration? Is Google already sentient”

“How is it even possible for every model to release benchmark results where they are #1 in 75% of categories?”

“Only theory is team wanted this out before perf/promo reviews to kick it over the line and then its not their problem”

“Gemini is so far behind that it is effectively useless compared to Claude.”

“Gemini not beating the "can't release a model" allegations”

“Why announce this if it’s not available yet? Why not at least announce when it will be released to the public? None of the other AI labs do this. Really frustrating.”

“At this point I just think they are benchmaxxing and all talk and no action.”

“This would explain why benchmarks are seemingly meaningless.”

“congratulations to people who work at google though on being allowed to use a model from 2026”

“Our departmental staffing has been cut down to 2/3 or 1/2 its historic level depending on when you measure from and we’ve been told basically never to expect another tenure line, but i can have free chatGPT to go with”

“Like not only did google force "upgrade" my voice control from Assistant to Gemini, half the time when I tell it to play a song in the car it tells me it can't or gives me a lecture on safe”

“While Gemini 4 has performed well on benchmarks the industry uses to gauge model efficacy, it does less well when employees actually put it to work”

“The enshittification of Google Maps continues. I just tried to find a locksmith along my route home, and the new Gemini button, after thinking for 30 seconds, suggested a 12 km, 1 hour detour all the way across Dublin”

“AI seems more distributed amongst neoclouds and traditional hyperscalers, FAANG and startups, GPUs and ASICs than it did this time a year ago. Nobody has a moat.”

“Google is getting punished for not letting the models enter an echo chamber and go faster than humanly possible.”

“Google is not letting outside people use their models because if they did they would have to scale up their TPU production faster than they can manage”

“Gemini 4 Argon is launching first in a limited capacity so Google can make sure it’s not misaligned. www.theverge.com/tech/1002980...”

“some Google employees say Gemini 4 performs well on benchmarks but struggles with some real-world coding tasks; Google disputes that characterization”

“Gemini doing reasonably well on the Hacking bench”

“I also want us to ask who gets left outside the gate, what happens to independent research, and whether safety controls track abuse rather than status!”

“Are they real or just benchmaxxed?”

“Gemini Argon takes more tokens to be able to deliver the same level of intelligence as Opus 5.5 (high; not max) so it would be slower and more expensive than Opus if not for the temporary 50% discount”

“Gemini 4 Argon tops OpenAI and Anthropic on most benchmarks, especially knowledge work, but its coding scores are mixed and access is limited.”

“Curious if it'll be actually useful for coding but also wonder if this model will maintain Gemini's track record of being frontier in multimodal environments.”

“only thing I was wrong about i I expected their model to be much stronger”