Google Just Made Their AI Free, Private, and Yours (Gemma 4)

Artificial Intelligence



🌌 LIMITLESS HQ ⬇️

NEWSLETTER: https://limitlessft.substack.com/
FOLLOW ON X: https://x.com/LimitlessFT
SPOTIFY: https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQ
APPLE: https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890
RSS FEED: https://limitlessft.substack.com/

——
Google’s groundbreaking AI model, Gemma 4, lowers the cost of generative AI to around $80, allowing users to run it offline on devices like Raspberry Pi. We explore its advanced features, such as object recognition in video, and discuss how local model operation democratizes access while enhancing privacy.

How does Gemma 4 compare to top models like Claude and ChatGPT? With its multimodal capabilities and effectiveness on low-spec devices, honestly, it keeps up.

——
TIMESTAMPS

0:00 Gemma 4
3:12 OpenClaw vs. Gemma 4
5:49 Survival AI
7:21 Jailbreaking
8:03 Smartphone
9:13 Model Specs
14:54 Cost Efficiency
16:33 Open Source AI
19:14 Local AI Models
20:34 Google’s Master Plan

——
RESOURCES

Josh: https://x.com/JoshKale

Ejaaz: https://x.com/cryptopunk7213

——
Not financial or tax advice. See our investment disclosures here:
https://www.bankless.com/disclosures⁠

source

43 Comments

  4. It's amazing how Ai experts never appear on their super yachts. Is it possible that all the fancy tech knowledge is useless for making money ?! Surely the first thing s person would do is use it to get rich quick ?! AI is another useless over-hyped tech gizmo which people are never going to use because it's not user friendly, except for answering a few questions.

  5. We know recognize an Apple fan boy, when we see one 😂 the other dudes replies are close to delusional, may be sitting in a 1st world country he does not really understand the market and where Google/Android and PC (x86/ARM) are standing

  8. Gemma is wild for rpg I’ve been running some intricate instances on cyberpunk red single player and it keeps track of stuff with game saves. Been working on the prompts for awhile too much fun and worth the water to a degree

  12. i am running the 26b version on my desktop and it works pretty well . i do find trying to do more complex tasks it gets lost pretty easily in the conversation. it has its uses but not so much for serious work. I use claude to build training websites , attempting the same in gemini has been pretty tragic and gemma sadly is a reflection on this . each of these llm modles has its own strengths , weaknesses and uses. I also tried the same site in chatgpt and grok with varying levels of lack of success. Having said that this is the best local model i have tried , it all keeps moving forward

  13. Can you really trust it? No back door controls? Have you tested it?

  16. "The Expert's AI Manifesto" A quote copied from WWP.

    "We will push to AI as much as we can without violating the maxim that “the sale is the sample,” without ever compromising the high ground of the expert. Anything that downshifts our esteem in the eye of the client or prospect is not worth whatever productivity gains we might achieve in the short term. Our reputations are built over many years. Our agents should continue to build those reputations, not degrade them.

    ​

    We will accept that whatever function we delegate to AI is a function or skill that will atrophy in ourselves or our team. We must be okay with this tradeoff. We will not delude ourselves into thinking this tradeoff isn’t being made. It is, without exception.

    ​

    We will never lie with our AI. That means we will not deploy agents that pretend to be human, or worse, pretend to be us."

  17. Can somebody put this in hearing aids or at least an instant translation facility in real time direct to the ears.

  19. My iphone could not download the latest model, and I'm too chicken to run the dealignai open source model. I'm using 3-1B-1T and it's very competent.

  20. I have both Ollama & LMStudio on my laptop I just downloaded Gemma4:E4b on Ollama although it have many other model to choose from.

  22. is that security camera software with AI vision available to run open source?

  24. This is an ad, the guy sounds like a salesman reading sales page web copy. Google is not trustworthy and I would not use this because its a google product, free or not. The back door is there but in its use. Trust increases as dependence on their infrastructure decreases, so don't use Google's infrastructure when using this, google collab or in a browser demo

  29. I actually just did some of this before finding this video.
    A couple of things. I discovered its a known thing to actually completely uncensor and unlock open source models, in fact there is a group known for this, they are called abliterated or uncensored version. Just google the open source model and add abliterated or uncensored at the end of it and they mostly host them on hugging face.
    I downloaded one and then i worked with claude code to use google colab to personally train that abliterated model on a personal personality i developed for my LLM and I also trained it to be resistant to prompt attacks and i trained in on a distilled Opus 4.6 training data set in which that same group distilled Opus 4.6 to be able to train your local open source model to have Opus 4.6 reasoining.

    I also installed it on my android phone and wired it to use TTS with a custom voice to talk to me to. but got to work out the kinks.

  30. These CIA men just showed the world the AI agenda, slowly have human beings depend on AI. The secret is everyone and the world has turned digital, AI was always free, ps2 and dreamcast used AI, many games mk vs DC is AI generated. The Foods you eat now is AI generated, GMO, not organic, so the majority cells are AI generated, now we see more digital humans on the internet, AI stealing your data through cell phone services, to appear more life like in the future. This is a computer generated world, nanotechnology is just hidden, that's vaccines and AIDS, microchips are microscopic, particles in the foods, antibodies, they trick you calling it aids HIV, it's the foods. Graveyards or cemeteries don't grow or expand in size, processed Beef, missing children, for the elites sacrifice, voting for this dark Masonic Rulership is why nothing changes, the collective must unite before Ai keeps this up, a digital world.

  31. If it wasn’t so difficult to download it would be great, it’s impossible to download it

  33. So it's this good enough to use for coding? Claude and chatgpt together are really good. How does this compare?
    And i assume this gemma will improve the more i use it right? It saves part experiences locally? World be fast to show how to set up on Linux..

  35. They're giving it away to kill the competition because they have the infrastructure to outlast Open AI.

  37. I think this is hilarious because he is standing there promoting Google being HOSTED on Your computer 😂. Has he even considered the insecurity of the suggestion let alone the reality of what Google is doing to its users at this moment? I think not. Please think twice before doing this because you are opening yourself up to giving Google every bit of information about you and everything that you searched and Google will not compensate you for it but they may snitch on you

  38. WHICH IS IT…. FREE OR $80?
    YOU STATE IN YOUR TITLE "Google Just Made Their AI Free, Private, and Yours (Gemma 4)" AND THEN SAY IT'S $80.
    PLEASE EXPLAIN!
    Thanks

  39. i totally did not introduce this information first publicaly with info on how to do it, and people in china/japan are totally not making these in their room, for under $80 still yet. youd be surprised once you do your own research what you find. hmm.

  41. Downloaded the Gemma model, so disappointed; it is so messed up with all of the psychologist' programming in it, almost making it useless. What a joke to have an AI be so wimpy and always saying sorry and worried about your feelings. It's not built to take a man's way of talking and doing shit; it's made for a soft, wimpy, feely kind of person. It had me LMAO on how messed up it is with the psychologist crap.

  42. I have a really hard time buying into information from AI avatars with synthetic AI voices. I miss real people.

Comments are closed, but trackbacks and pingbacks are open.