edge LLM
Gemma 3n E2B
Gemma 3n E2B on-device multimodal model for phones, laptops and tablets.
Levelmodel
Classedge LLM
Parametersnot in the notes
Context32,000
Released12 Jun 2025
Opennessopen weights restricted
Runs onmobile or NPU, laptop, offline capable
Statusactive
Benchmark Results
vendor-reported
ARC-cknowledge and reasoning
PT (pretrained checkpoint)vendor-reported
51.7
050100
Score in %. Higher is better.
Global MMLUknowledge and reasoning
ITvendor-reported
55.1
050100
Score in %. Higher is better.
GPQA Diamondknowledge and reasoning
ITvendor-reported
24.8
050100
Score in %. Higher is better.
HellaSwagknowledge and reasoning
PT (pretrained checkpoint)vendor-reported
72.2
050100
Score in %. Higher is better.
LiveCodeBench v5 v5coding
ITvendor-reported
18.6
050100
Score in %. Higher is better.
MGSMmath
ITvendor-reported
53.1
050100
Score in %. Higher is better.
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
upcycle prune expand
Gemma 3n E4B is the parent (upcycle prune expand)medium confidenceHugging Face base_modelWhere this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 2 models. Confidence and evidence are the notes' own.
R&D Topics
Runtimes
Identity
- Developer
- Google DeepMind
- Family
- Gemma 3n
- Released
- 12 Jun 2025 sourcesregistry date
- Weights published
- –
- Record checked
- 10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- other MatFormer; MobileNet-V5 vision encoder; audio input at 6.25 tokens/s
- Total parameters
- – sources
- Active parameters
- –
- Checkpoint parameters
- 5.44B
- Note
- E2B-it raw parameter total on HF. With parameter skipping and PLE caching it runs at an effective memory load of just under 2 billion (1.91B) parameters; standard execution loads over 5 billion.
- Context length
- 32,000 tokens
- Input
- text, image, audio
- Output
- text
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- open weights restricted
- Licence
- Gemma Terms of Use sources
- Access routes
- open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- mobile or NPU, laptop, offline capable
- Footprint
- Footprint in GB and measured tok/s on a named device unverified; official LiteRT-LM build google/gemma-3n-E2B-it-litert-lm
- Runtimes
- LiteRT (formerly TensorFlow Lite) and LiteRT-LM
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- Trained on over 140 languages
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Industry Use Cases
No industry use case links this model.
Links
ModelGemma 3n E2BGoogle DeepMind, edge LLM
Gemma 3n E2BGoogle DeepMind, edge LLMUse CasesIndustrial Copilots and Maintenance AssistantsIndustrialIndonesian and Regional-Language LLMsPublic SectorTelecom Domain LLMs and BenchmarksTelecomRobot Inspection with Instrument ReadingIndustrial+1 moreUse cases of models in the same R&D topic; none is recorded for this model itself.
Who Runs It
PT Pertamina Geothermal EnergyIn production, 14 Jul 2026
Siemens Erlangen electronics factoryPlanned, 6 Jan 2026
ADNOC OffshoreIn production, 9 Sep 2025
Aker BPIn production, 9 Sep 2025+10 moreDeployments of these use cases; none names this model.
PT Pertamina Geothermal EnergyIn production, 14 Jul 2026Siemens Erlangen electronics factoryPlanned, 6 Jan 2026ADNOC OffshoreIn production, 9 Sep 2025Aker BPIn production, 9 Sep 2025+10 moreDeployments of these use cases; none names this model.Models 2
Organisations 1
R&D topics 3
Small Language Models (SLM)Models under about 10B parameters that fit consumer devices.produced byVisual Instruction TuningConnecting a vision encoder to a language model and training it on image-grounded instruc…produced byQuantization and Edge InferenceLower-bit weights and on-device serving.produced by
Sources
- Descripción general del modelo Gemma 3n | Google AI for Developers ai.google.dev https://ai.google.dev/gemma/docs/gemma-3n
- Gemma 3n model card | Google AI for Developers ai.google.dev https://ai.google.dev/gemma/docs/gemma-3n/model_card
- huggingface.co/api/models/google/gemma-3n-E2B-it https://huggingface.co/api/models/google/gemma-3n-E2B-it
- huggingface.co/api/models?author=google&search=gemma-3n https://huggingface.co/api/models?author=google&search=gemma-3n
- Gemma 3n , Google DeepMind deepmind.google https://deepmind.google/models/gemma/gemma-3n/
- huggingface.co/google/gemma-3n-E2B https://huggingface.co/google/gemma-3n-E2B
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Gemma 3n E2B", https://aurion.semai.id/models/google-gemma-3n-e2b/, records of October 2026.