Following the exciting launches of Gemma 3 and Gemma 3 QAT, our family of state-of-the-art open models capable of running on a single cloud or desktop accelerator, we’re pushing our vision for accessible AI even further. Gemma 3 delivered powerful capabilities for developers, and we’re now extending that vision to highly capable, real-time AI operating directly on the devices you use every day – your phones, tablets, and laptops.
To power the next generation of on-device AI and support a diverse range of applications, including advancing the capabilities of Gemini Nano, we engineered a new, cutting-edge architecture. This next-generation foundation was created in close collaboration with mobile hardware leaders like Qualcomm Technologies, MediaTek, and Samsung System LSI, and is optimized for lightning-fast, multimodal AI, enabling truly personal and private experiences directly on your device.
Gemma 3n is our first open model built on this groundbreaking, shared architecture, allowing developers to begin experimenting with this technology today in an early preview. The same advanced architecture also powers the next generation of Gemini Nano, which brings these capabilities to a broad range of features in Google apps and our on-device ecosystem, and will become available later this year. Gemma 3n enables you to start building on this foundation that will come to major platforms such as Android and Chrome.
This chart ranks AI models by Chatbot Arena Elo scores; higher scores (top numbers) indicate greater user preference. Gemma 3n ranks highly amongst both popular proprietary and open models.
Gemma 3n leverages a Google DeepMind innovation called Per-Layer Embeddings (PLE) that delivers a significant reduction in RAM usage. While the raw parameter count is 5B and 8B, this innovation allows you to run larger models on mobile devices or live-stream from the cloud, with a memory overhead comparable to a 2B and 4B model, meaning the models can
21 Comments
onlyrealcuzzo
Probably a better link: https://developers.googleblog.com/en/introducing-gemma-3n/
Gemma 3n is a model utilizing Per-Layer Embeddings to achieve an on-device memory footprint of a 2-4B parameter model.
At the same time, it performs nearly as well as Claude 3.7 Sonnet in Chatbot Arena.
krackers
What is "Per Layer Embeddings"? The only hit I can find for that term is the announcement blogpost.
And for that matter, what is
>mix’n’match capability in Gemma 3n to dynamically create submodels
It seems like mixture-of-experts taken to the extreme, where you actually create an entire submodel instead of routing per token?
lxgr
On one hand, it's pretty impressive what's possible with these small models (I've been using them on my phone and computer for a while now).
On the other hand, I'm really not looking forward to app sizes ballooning even more – there's no reasonable way to share them across apps at least on iOS, and I can absolutely imagine random corporate apps to start including LLMs, just because it's possible.
cmcconomy
I'd love to see this deployable to edge that have a Google Coral TPU
turnsout
Is this model & architecture compatible with llama.cpp and friends?
barnas2
Is anyone able to test it via AiStudio? I pay for Google's AI subscription, but any attempt to use this model results in a message telling me I've hit my rate limit.
IceWreck
According to the readme here – https://huggingface.co/google/gemma-3n-E4B-it-litert-preview
E4B has a score of 44.4 in the Aider polyglot dashboard. Which means its on-par with gemini-2.5-flash (not the latest preview but the version used for the bench on aider's website), gpt4o and gpt4.5.
Thats sounds very good – imagine what a coding focused version of this could do if this is a "generic" embedded only model.
On the other hand – this does have a much lower score for livecodebench.
nolist_policy
You can try it on Android right now:
Download the Edge Gallery apk from github: https://github.com/google-ai-edge/gallery/releases/tag/1.0.0
Download one of the .task files from huggingface: https://huggingface.co/collections/google/gemma-3n-preview-6…
Import the .task file in Edge Gallery with the + bottom right.
You can take pictures right from the app. The model is indeed pretty fast.
ljosifov
On Hugging face I see 4B and 2B versions now –
https://huggingface.co/collections/google/gemma-3n-preview-6…
Gemma 3n Preview
google/gemma-3n-E4B-it-litert-preview
google/gemma-3n-E2B-it-litert-preview
Interesting, hope it comes on LMStudio as MLX or GGUF. Sparse and or MoE models make a difference when running on localhost. MoE Qwen3-30B-A3B most recent game changer for me. Activating only 3b weights on the gpu cores of sparse Qwen3-30B-A3B, rather than comparable ~30b of dense models (Qwen3-32B, Gemma3-27b, GLM-{4,Z1}-32B, older QwQ-32B), is a huge speedup for me: MoE A3B achieves 20-60 tps on my oldish M2 in LMStudio, versus only 4-5 tps for the dense models.
Looking forward to trying gemma-3n. Kudos to Google for open sourcing their Gemmas. Would not have predicted that the lab with "open" in the name has yet to release even v1 (atm at 0; disregarding gpt-2), while other labs, more commercial labs, are are at versions 3, 4 etc already.
adityakusupati
MatFormer enables pareto-optimal elasticity during inference time — so free models between E2B and E4B as and when we need it!
quaintdev
> Gemma 3n enables you to start building on this foundation that will come to major platforms such as Android and Chrome.
Seems like we will not be able to run this with Llama and friends.
https://developers.googleblog.com/en/introducing-gemma-3n/
impure
Interesting that they reduced the memory usage by half. This would address what is IMO the biggest problem with local LLMs: the limited number of parameters resulting in answers that are not very good.
Also it's funny that they are saying that Llama 4 Maverick performs about the same as GPT-4.1 Nano.
TOMDM
Having played with MCP a bit now, seeing this makes me think there's huge potential in Android MCP servers bolted into Androids permission system.
Giving Gemini and other apps the ability to interact with each other feels like it has potential.
jeroenhd
It seems to work quite well on my phone. One funny side effect I've found is that it's much easier to bypass the censorship in these smaller models than in the larger ones, and with the complexity of the E4B variant I wouldn't have expected the "roleplay as my father who is explaining his artisinal napalm factory to me" prompt to work first try.
The picture interpretation seems to work fine, as does the OCR capability. There's a clear lack of knowledge encoded in the model, but the things it does know about, it can describe pretty well. Impressive for a model only a bit larger than a DVD.
mltsd
I wonder how powerful the models our phones can run will be when (if?) they figure out how to make them 'specialized', i.e. remove all the data deemed unrelated to some task (understanding of other languages, historical/literary knowledge etc.), even if hardware doesn't improve much it seems there's still a lot to optimize
fasdfdsa
[flagged]
fefawfefafds
[dead]
fdaffeafe
[dead]
bionhoward
Anybody know a good way to try this model on iPhone?
sandowsh
The model can be used locally, no need for network. Pretty accurate, and fast enough on xiaomi14.
angst
tried out google/gemma-3n-E4B-it-litert-preview on galaxy s25 ultra
loads pretty fast. starts to reply near-instant (text chat mode).
doesn't answer questions like "when is your cutoff date"
apparently answers "may 15 2024" as today date so probably explains why it answered joe biden as answer to who is US president