Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.
Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.
As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.
Rules:
Rule 1 - No harassment or personal character attacks of community members. I.E no namecalling, no generalizing entire groups of people that make up our community, no baseless personal insults.
Rule 2 - No comparing artificial intelligence/machine learning models to cryptocurrency. I.E no comparing the usefulness of models to that of NFTs, no comparing the resource usage required to train a model is anything close to maintaining a blockchain/ mining for crypto, no implying its just a fad/bubble that will leave people with nothing of value when it burst.
Rule 3 - No comparing artificial intelligence/machine learning to simple text prediction algorithms. I.E statements such as "llms are basically just simple text predictions like what your phone keyboard autocorrect uses, and they're still using the same algorithms since <over 10 years ago>.
Rule 4 - No implying that models are devoid of purpose or potential for enriching peoples lives.
Probably because they play the same game as Mark Zuckerberg, the Chinese, to some degree OpenAI... They all release open-weights models.
They'll generate some hype for their company that way, so it's advertising. They build good will. They undercut the competition. Or make it clear how they outperform them. Maybe they get some more investor money if they do expand to the local models market. I bet there's a million reason why it makes sense from a business perspective.
It's called commoditizing your complements. It's a complex strategy which Google has done before with Android and Chrome. They're masters at it.
The same reason they made Android open source. To look good (Hey we aren't like the other guys. We support open source!) and to bring in outside ideas, input, and work until they close off every avenue and it's open in name only.
Yeah, I'm thinking they want phone devs to start adding their models to android and create a similar situation where it becomes so integrated that it's almost impossible to get rid off. Like they did with chromium, where most browser are now stuck with it and they can change the rules whenever they want.
These gemma models are small enough to run on your laptop or phone. They'll be bundling them with phones and Chromebooks anyway.
Might as well get some goodwill and let the horses run free before someone extracts them from the edge device anyway.
It can give them some plausible deniability if they just ship them off as tech-previews. I'm thinking for when that taking kitty app starts feeding into your delusions and telling you to hurt people.
Prophetic since a week later people found Google pushing AI secretly on Google Chrome. Hexavalent chrome is banned, why can't chrome be banned?
I'll further prophesize that we'll start seeing mixed on-device and cloud calling. Cloud for heavy thinking is probably in the books right now.
Next week your local tiny gemma4 will be feeding the cloud models with predicted tokens to speed up and reduce work for gemini. It only has to get it right 66% of the time for a 2x speed-up.
A lot has been said, but to add to the list I'd say it gives them access to quite a large pool of free testers.
LLM architectures and optimization techniques change rapidly and by releasing open-weight models a lot of enthusiasts will evaluate new models for free, help implement support in inference engines, catch bugs etc. (and in turn, ofc, get a new model to run for free, so it's at least somewhat symbiotic).
We have at least seen this quite obviously when Alibaba released Qwen3-Next, which was a somewhat undertrained but still useful model which introduced the architecture that their latest models now use "in production" (also their paid "Max" models).
Cause no one really cares about them and Google doesn't want to fade into obscurity in this AI race.
Random suspicion: It is correlated to how much compute US have. That's the only reason I can think of. I also wondered why they would do that when US tech are grabbing the global hardware. They'd like to keep control of inference in the cloud but are missing power thus lack compute. But they also benefit from a western AI ecosystem that have much more compute available. Pure guess, but maybe something like that ?
Marketing