This is just early fusion basically.

FAIR did this 2 years ago now: https://arxiv.org/abs/2405.09818

I've been waiting for something like this to be released since then.

The annoying thing is that chameleon was multi-modal out based on the same principles, but this model is just inputs... (I'm curious how they did pre-training without having multi-modal outputs as well. I wonder if they just chopped them off rather than support image output).

Embedded within that developer page is a good explainer of the encoder free architecture . https://newsletter.maartengrootendorst.com/p/a-visual-guide-...

> quantization

12b means 12G @ 8 bits/param (basically lossless) and 6G at 4 b/p (generally accepted 'pretty close' level). Not too bad?

But TBD how well the base model performs before thinking too much about quantization

Totally agree that it is "encoding" in the general sense, but I think they are referring to the lack of an "encoder" neural network.

The audio side is even more interesting, as it seems they totally got rid of positional embedding are just doing a single linear transform to match the LLM input dimension and that's it.

> Audio: We simplified audio processing even further. We removed the audio encoder entirely and projected the raw audio signal into the same dimensional space as text tokens.

One side-effect, is that the separate .mmproj file (Multi-Modal Projection encoder) is no longer needed, when using the model with llama.cpp etc.

Encoder free is huge for running on SBCs etc. often the encoding time is a significant fraction of generation time if you are using a VLM as a all purpose vision model

It actually works well because unlike encoders, the latent space is trained on that initial layer so it “knows” what to do with that sparse density. I’ve been using gemma4-12b with Flux2 and its ability to reason on visual input is pretty good. That said, each model is good in their own ways so YMMV but overall, it’s about as solid as Qwen just with a more advanced architecture.

I think the idea is that the model is seeing embeddings that map directly to underlying pixel data, rather than being fed semantically rich embeddings from an encoder model which itself had seen the raw pixel data.

Well its a real simple encoder I guess

> That's technically encoding

Isn't that just projecting the patches into the d_model size vectors that the models takes?

>I am assuming that involves of quantization

12B model in 16GB seems very reasonable to me, int8 is top quality for running models.

[dead]

A big part of the frontier labs abilities to charge 80% gross margins on inference is having the cornered resource of frontier models.

If that inference becomes popular and valuable enough that those companies make billions of dollars in profit, those companies could use that profit to fund the building of alternative products and platforms that dis-intermediate google's relationship with the customer.

Google already has an 80% gross margin business, the biggest one in the world. Everybody wants a slice of it.

By offering frontier inference closer to cost and open-sourcing everything that's sub-frontier, they're commoditizing frontier labs' models, which inhibits their ability to durably make high gross margins on inference.

It's a strategic play.

This won't replace commercially viable, revenue generating alternatives of their own devising, but it does enable development activity and initiate conversations with enterprises who start with this model but want to do slightly more.

That's my experience right now... my company is all in on a plethora of platform products. Also, Microsoft just yesterday said their goal was "Unmetered intelligence". There's a lot of things that can be enabled by small local models, and those things are part of stacks that can generate revenue in other layers.

Android and Chrome need on-device AI capabilities. Google can't lock down those weights like it can with server-side ML.

So it's easier to just release those models as open source and make it official, since someone would inevitably hack the weights out anyway.

Google is one of the few verticalized options in AI: Data, models, cloud services, low-level silicon (TPUs), internal use cases, retail use cases, B2B uses, distribution (browser & mobile), etc.

They rise with the tide of AI adoption. But they gain ground if people opt into Google solutions. And any token sent to a Google model (free or paid) actively punishes their competitors that are then required to spend vast sums to remain bleeding edge.

As long as Chinese firms are releasing good open models I imagine there isn't a huge downside for Google to release state of the art small models to compete in the "free" space.

If you're an AI lab, you definitely want research teams in this space - as this is where you can most easily iterate and make improvements which you'll then bake into larger, frontier models.

The question is: do you want to release your models, or use them purely for R&D?

Since everyone else is already releasing models of similar qualities, it's hard to say you're shooting yourself in the foot if you join the chorus.

The added cannibalization of releasing them is effectively zero, so the reputational benefits are likely to be worth it.

It's to destroy possible footholds for competitors and prevent them from making money in segments that Google doesn't care too much about, but can trivially commoditize.

Neutering OpenAI and Anthropic would be my guess. Commoditized LLMs won't hurt Google nearly as much as it hurts the LLM-only companies, and so accelerating the inevitable just helps knock out potential future competition in areas where Google -does- make a lot of money now.

Maybe they are hedging against a future where local models are just as good as cloud models? Or maybe they can go the Taalas route and start hardcoding Gemma on a chip and hardware manufacturers can use it for local private AI.

Isn't Apple about to license some variation of this from google for on-device AI? Maybe it’s their sales pitch to Apple and then they will lock it down.

They're trying to capture the segment of the market that wants to control the model, with the intent of getting you to run them on Vertex.

My guess is testing for Apple’s Siri replacement and partnership but that’s a total SWAG

Marketing + Pro Serv if I had to take a guess.

Evangelism for AI. Google is one of the big AI providers.

Eventually the local model is not enough, and you'll upgrade to the big ones.

edge compute

Gemma overtakes and kills real open-source AI projects, pushing people who would support them towards enterprises like Google

Google's MO since always has been to release great products or services for free, position themselves high and then abandon them or just find uses for Enterprise sales.

I'm pretty sure they are doing it because they get some research experience by shrinking and improving these models, and because they know that by doing this they get some good PR among the dev community.

Since ollama has diverged from llama.cpp, it will take a bit of time for ollama to support multi-modality. If you're using plain llama.cpp it looks like a PR has already merged for this model with vision and audio support:

https://github.com/ggml-org/llama.cpp/pull/24077

MLX is quite literally macOS-specific technology, for other platforms you want non-MLX.

I was sure "MLX" stood for "Metal-something-something" but can't find any reference to that somehow, anywho, "Metal" is hardware-accelerated graphics on Apple platforms FWIW.

Edit: about the actual release on Ollama, if you're on non-Apple hardware you probably want the NVFP4 variant ("gemma4:12b-nvfp4") which was uploaded 45 minutes ago, especially if you're with a recent nvidia GPU.

There's a CUDA backend for MLX now. Not sure about the maturity.

MLX is Apple’s own machine learning framework, designed for Apple Silicon: https://opensource.apple.com/projects/mlx/

The un-quantized MoE outperforms it.

But between same (V)RAM requirement 4 bit 26B-A3B and 8 bit 12B it's unclear which one will win, especially given one is MoE and the other dense.

All the launch benchmarks are at 16 bit.

"Small" models are the ones I can run myself on my own terms. LLMs aren't useful enough for me to justify spending hundreds of euros on a GPU with 16GB VRAM or something, and that's assuming I have the rest of the desktop just laying around. Back when I checked (before the RAM price hike), these models weren't meaningfully better than 4-8GB ones anyway, you'd have to go for the top tier cards at 24 or 32 GB iirc to get something vaguely in the direction of the SaaS versions, and that was absolutely out of my budget. Even if that changed, so have hardware prices so it'd probably still work out the same

I've yet to see someone answer a question like this with a decent, useful answer.

I dunno, feels a bit unfair to companies that actually do FOSS releases (Gemma 4 being released under Apache 2.0 license) to compare them to a company that never done any FOSS releases, and mostly done proprietary "available to download" releases.

Every other Google model I have tried felt very weak compared to qwen models. I dont have a ton of use case for multimodal though, so its very possible this is a fantastic multimodal model.

IDK this model release is a bit disappointing considering the community has been chomping at the bit for the 124ba4b model. There was some leaked info about it but people suspect it was not released because it was too close to gemini flash in performance.

This comment has me a bit confused.

Consumers were complaining about the standard 8GB with the early 2020 refresh of MacBook Pros, many OSes ago. Sure, it might be workable for many tasks (as evidenced by the recent sales of the MacBook Neo), but users with a mere 8GB shouldn't have expectations of LLM performance. Even 16GB feels like a stretch.

There's a comparison of all the Gemma 4 models (+ Gemma 3 27B) on the Huggingface model card: https://huggingface.co/google/gemma-4-12B-it#benchmark-resul...

That's what I want to know too. A smarter E4B that's happy in opencode would be a good selfhosted model for me