Ministral 14B
Trinity Nano Preview
12B
- Context
- 33Ktokens
- Released
- 1y ago
- Size
- 7.8Bparameters
Efficient on-device Gemma 3n model for text, image, and audio understanding
- Multimodal model
- Open weights · 7.8B
- Accepts images, audio
How it compares
Size class
Small7.8BAn open-weight model of 5 to 25 billion parameters, the size that runs on a single consumer GPU.
Open models of a similar size from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inFamily tree
Models trained from one another, as their makers describe it.
Gemma 3n E4B ITGoogle DeepMind · 2025
Additional source evidence
- Source model
- google/gemma-3n-e4b-it
- Source family
- gemma
- Last updated
- 2025-06-26
- Knowledge cutoff
- 2024-06
- Model context
- 33K
- Model output limit
- 33K
- Open weights
- Yes
- File attachments
- Yes
- Reasoning
- No
- Tool calling
- No
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "google/gemma-3n-e4b-it",
"name": "Gemma 3n E4B IT",
"description": "Efficient on-device Gemma 3n model for text, image, and audio understanding",
"family": "gemma",
"last_updated": "2025-06-26",
"knowledge": "2024-06",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"limit": {
"context": 32768,
"output": 32768
},
"release_date": "2025-06-26",
"open_weights": true,
"license": "Gemma Terms of Use",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/google/gemma-3n-E4B-it"
}
]
}Text provenance
Gemini app/webGoogle SynthID text watermarking
Google DeepMind says text generated in the Gemini app and web experience is watermarked with SynthID by adjusting token probabilities.
Official explanationFrequently asked questions
Running locally
Gemma 3n E4B IT has 7.8B parameters and needs an estimated 6 GB of VRAM at Q4, 10.1 GB at Q8 and 18.6 GB at FP16. All figures are estimates, consisting of weights plus about 15% and 0.5 GB for a short context.
Yes. It fits on a 24 GB GPU at FP16 (estimated 18.6 GB), on 16 GB and 12 GB GPUs at Q8 (estimated 10.1 GB) and on an 8 GB GPU at Q4 (estimated 6 GB).
Yes, at Q4 with an estimated 6 GB of VRAM. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits is 8 GB.
Privacy and data
Yes. Gemma 3n E4B IT has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window of Gemma 3n E4B IT is 33K tokens (32,768). The maximum output is also 33K tokens (32,768).
Gemma 3n E4B IT accepts text, image and audio inputs. Video and PDFs are not listed among its input modalities.
No. Gemma 3n E4B IT does not support tool calling (function calling).
No. Gemma 3n E4B IT does not have a reasoning or thinking mode.
The knowledge cutoff of Gemma 3n E4B IT is 2024-06.
Quality and benchmarks
Gemma 3n E4B IT ranks 272 on Arena Text, which is Arena's own rank, with a score of 1317.
The closest alternatives, matched on model kind and release date with no comparable price, are ERNIE 4.5 0.3B by Baidu, ERNIE 4.5 VL 28B A3B by Baidu and DiffuCoder 7B by Apple.
ERNIE 4.5 0.3BGemma 3n E4B IT vs ERNIE 4.5 0.3BERNIE 4.5 VL 28B A3BGemma 3n E4B IT vs ERNIE 4.5 VL 28B A3BDiffuCoder 7BGemma 3n E4B IT vs DiffuCoder 7B
Versions and lineage
Both models have a context window of 33K tokens (32,768) and were both released on June 26, 2025, so no difference between them is listed in the facts.
Licensing and access
Gemma 3n E4B IT is open weights rather than fully open source, licensed under the Gemma Terms of Use.
The Gemma 3n E4B IT weights can be downloaded from the Hugging Face repository google/gemma-3n-E4B-it.
Basics
Gemma 3n E4B IT was made by Google DeepMind and released on June 26, 2025.
Yes, Gemma 3n E4B IT has open weights.
Generated from Vertical Tabs data, updated November 11, 2026.
Discussion
No comments yet
Be the first to share your thoughts.