Earlier quoted context omitted.
it is. itll still not create AGI without some breakthrough in instruction vs data separation of concerns
In what way do we not already have AGI?
in all of them
Our eighth generation TPUs: two chips for the agentic era
201–210 of 240 posts
Re: Our eighth generation TPUs: two chips for the agentic era
#202At this point, when you are doing big AI you basically have to buy it from NVidia or rent it from Google. And Google can design their chips and engine and systems in a whole-datacenter context, centralizing some aspects that are impossible for chip vendors to centralize, so I suspect that when things get really big, Google's systems will always be more cost-efficient. (disclosure: I am long GOOG, for this and a few o…
Don't build your castle in someone else's kingdom. Buying from nvidia is the only real option and even that is not optimal.
> Don't build your castle in someone else's kingdom.
would like to know about the scrape content of these castles /j
Re: Our eighth generation TPUs: two chips for the agentic era
#203At this point, when you are doing big AI you basically have to buy it from NVidia or rent it from Google. And Google can design their chips and engine and systems in a whole-datacenter context, centralizing some aspects that are impossible for chip vendors to centralize, so I suspect that when things get really big, Google's systems will always be more cost-efficient. (disclosure: I am long GOOG, for this and a few o…
I think this is a narrow view. Aws and azure build their own data centers and partner closely with Nvidia and build their own silicon too. TPUS are non standard, no one else can run them - Nvidia build on fabrics and technologies well under and well integrated for a long time (mellanox etc) and clearly work very closely with the aws and azure hardware and data center build teams. I’d not bet that Google can do things better than everyone else - that’s certainly something Googlers always believe about themselves but it’s not the case that you can’t build a best of breed that meets or exceeds total in house builds.
Re: Our eighth generation TPUs: two chips for the agentic era
#204I already felt that gemini 3 proved what is possible if you train a model for efficiency. If I had to guess the pro and flash variants are 5x to 10x smaller than opus and gpt-5 class models. They produce drastically lower amount of tokens to solve a problem, but they haven't seem to have put enough effort into refinining their reasoning and execution as they produce broken toolcalls and generally struggle with 'agent…
Am I tripping or is this an AI reply? Like it barely has anything to do with the article other than both are related to AI
Google uses these chips to create gemini, I simply used this as an excuse to rant and predict the future.
Re: Our eighth generation TPUs: two chips for the agentic era
#205I already felt that gemini 3 proved what is possible if you train a model for efficiency. If I had to guess the pro and flash variants are 5x to 10x smaller than opus and gpt-5 class models. They produce drastically lower amount of tokens to solve a problem, but they haven't seem to have put enough effort into refinining their reasoning and execution as they produce broken toolcalls and generally struggle with 'agent…
> a model that will be an entire generation beyond SOTA That model would then be SOTA. Tautologically you can't be better than SOTA
SOTA at that time*
Re: Our eighth generation TPUs: two chips for the agentic era
#206I already felt that gemini 3 proved what is possible if you train a model for efficiency. If I had to guess the pro and flash variants are 5x to 10x smaller than opus and gpt-5 class models. They produce drastically lower amount of tokens to solve a problem, but they haven't seem to have put enough effort into refinining their reasoning and execution as they produce broken toolcalls and generally struggle with 'agent…
I really wonder what I’m missing with Gemini. It’s a second rate model for me at best. I find it okay (not great) at collecting information and completely useless at agentic tasks. It’s like it’s always drunk. When the Claude credits expire in Antigravity, I’m done for the day. > They produce drastically lower amount of tokens to solve a problem I LOLed at this because I of the constant death loops that don’t even so…
Yah it doesn't even make sense how they got through their benchmarks without death loops. Gemini-cli even has a hotfix to break the model of such death-loops. But if you were to ignore this bug/quirk that will be fixed in the next patch release my point still stands.
Re: Our eighth generation TPUs: two chips for the agentic era
#207I already felt that gemini 3 proved what is possible if you train a model for efficiency. If I had to guess the pro and flash variants are 5x to 10x smaller than opus and gpt-5 class models. They produce drastically lower amount of tokens to solve a problem, but they haven't seem to have put enough effort into refinining their reasoning and execution as they produce broken toolcalls and generally struggle with 'agent…
Interesting mix of words: "I felt" -> "proved" -> "guess". One of those is not like the others!
I guess I felt pretty uncertain that day which proved that a lack of sleep is bad for your mental cognition.
Re: Our eighth generation TPUs: two chips for the agentic era
#208Earlier quoted context omitted.
> If I had to guess the pro and flash variants are 5x to 10x smaller than opus and gpt-5 class models. I really doubt it, especially Pro. If anything I wouldn't be surprised if their hardware lets them run bigger models more cheaply and quickly than the others. Pro is probably smaller than GPT 5.4 and Opus 4.6 (looks like 4.7 decreased in size), but 5x seems way too much. IMO Gemini 3 Pro is the most "intelligent" in…
Regarding Anthropic, they used to make best multilingual and generalist models, it's their policy thing, not a capability issue. Claude 3 was best at this, including dead and low-resource languages. Neither modern Claude nor Gemini are remotely close to what Claude 3 was capable of (e.g. zero-shot writing styles). Anthropic basically reversed their "character training" policy and started optimizing their models for c…
This always bothers me because models will almost never see text that is mostly English with a little other language in training data (opposite happens of course) and certainly not in RL data. Why do they occasionally language switch?
Re: Our eighth generation TPUs: two chips for the agentic era
#209Earlier quoted context omitted.
As other people are saying here: the Gemini models are mostly terrible at tool use and long context management. And maybe not quite as good with finicky "detail" parts of coding generally. Where they excel is just total holistic _knowledge_ about the world. I don't like "talking" to it, because I kind of hate its tone, but I find Gemini generally extremely useful for research and analysis tasks and looking up informa…
People who say Gemini is bad at long contexts are so wrong. You can put whole 50,000 - 70,000 LOC codebase into Gemini 3.1 Pro context making it 800,000+ tokens, give it detailed task and ask for whole changed files back and it will execute it sometimes in one shot, sometimes in two. E.g depend on whatever stack you work with let you see all the errors at once so it can fix everything on single reply. Yes it will giv…
This! And with AI studio you get a couple of free calls per day (it has gotten less and less). I have had days where I would be able to get 100 USD worth of tokens from AI studio for free. 1m tokens in and great code out.
Re: Our eighth generation TPUs: two chips for the agentic era
#210Earlier quoted context omitted.
I've been trying Gemini Pro using their $20-ish Goole One subscription for a couple of months, and I also find it consistently does fewer web searches to verify information than say ChatGPT 5.4 Pro which I have through work. I was planning on comparing them on coding but I didn't get the Gemini VSCode add-in to work so yeah, no dice. The Android and web app is also riddled with bugs, including ones that makes you los…
I don't sweat sources and almost never check them. I usually prefer to manually check information after it's provided, to prevent the model from borking it's context trying to find sources that justify it's already computed output. Almost all the knowledge is already baked into the latent space of the model, so citing sources generally is a backwards process. I see it like going to the doctor and asking them to cite…
Well, I prefer it actually check datasheets so it doesn't go on a wild rabbit hunt to nowhere, since the capabilities it hallucinated for the chip in question doesn't exist.