SAVRN
Search Contact SAVRN

Open-weight model · Any to any

gemma-4-12B-it

by Google google/gemma-4-12B-it

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output.

Parameters12B
Context262,144
Weights23.9 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads2.7M

Runs On

What it takes to serve gemma-4-12B-it (12B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 23.9 GB 28.7 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 12.0 GB 14.4 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 6.0 GB 7.2 GB 1x MI300X (192 GB)
Vultr		 $1.85 1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.

SAVRN's Notes on gemma-4-12B-it

Text and images in, audio too on this size, text out, 140-plus languages, a 256K-token window: that is Google's instruction-tuned Gemma 4 12B. The footprint is 28.7 GB at 16-bit, 14.4 GB at 8-bit and 7.2 GB at 4-bit, all three on one MI300X at $1.85 per hour with 192 GB to work in, so the question is not which card but how much of it you leave for a 262,144-token context and its neighbors.

Nothing in Apache 2.0 stops you: commercial use, modification and redistribution, with license and NOTICE files kept and changes stated. It derives from google/gemma-4-12B, the pre-trained base, so start there for your own tuning, and read arXiv:2607.02770 first. The scores here, 78.8 on GPQA diamond among them, are model card numbers, not ours; run your own documents. No Index host prices it by the token; the comparison is the card hour.

SAVRN Research, 2026-09-18

Model Card

By Google, published under apache-2.0, revision 707f0a3b8a3c.

Hugging Face | GitHub | Launch Blog | Documentation | Technical Report
License: Apache 2.0 | Authors: Google DeepMind

[!Note] This model card is for the Gemma 4 12B Unified model, which is part of the Gemma 4 family of open models. Built with the same multimodal functionality as Gemma 4 E2B and E4B (text, audio, image, and video inputs), it brings native audio and vision understanding directly to local environments without the need for separate encoders. This unified approach to multimodality makes the model encoder-free, offering a deployment size that is perfect for consumer devices and streamlined local execution.

Read the full model card (3,506 words)

Configuration

Architecture
Gemma4UnifiedForConditionalGeneration
Context length (tokens)
262,144
Layers
48
Hidden size
3,840
Feed-forward size
15,360
Attention heads
16
Key/value heads
8
Head dimension
256
Vocabulary size
262,144
Sliding window (tokens)
1,024
Model type
gemma4_unified

Identity and Version

Repository
google/gemma-4-12B-it
Publisher
Google
Task
Any to any
Modality
Multimodal
Library
transformers
Parameters
12B parameters
Languages
Not stated by the source
Revision
707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7
First published
2026-05-23
Last updated
2026-07-20

Files and Weights

9 files, 24.0 GB in total. The weights are 1 file totalling 23.9 GB in safetensors.

Weights1 file · 23.9 GB
Configuration3 files · 6.1 KB
Tokenizer2 files · 32.2 MB
Documentation1 file · 28.5 KB
Other1 file · 18.7 KB
Repository1 file · 1.6 KB
Every file
FileTypeSizeSHA-256
model.safetensorsWeights23.9 GB 5a84cb313260
config.jsonConfiguration4.4 KB —
generation_config.jsonConfiguration260 B —
processor_config.jsonConfiguration1.4 KB —
README.mdDocumentation28.5 KB —
chat_template.jinjaOther18.7 KB —
.gitattributesRepository1.6 KB —
tokenizer.jsonTokenizer32.2 MB cc8d3a0ce364
tokenizer_config.jsonTokenizer3.1 KB —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
23.9 GB
Download from Google

Released by Google through Kaggle. Read the license.

Built From

Evaluations

Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.

BenchmarkConditionsResultReported byRevisionDate
Idavidrein/gpqa Task diamondMetric diamondComparison conditions not established 78.8 Model Card
Reported by a third party		 Evaluated revision not stated 2026-05-23
MMMU/MMMU_Pro Task mmmu_proMetric mmmu_proComparison conditions not established 69.1 Model Card
Reported by a third party		 Evaluated revision not stated 2026-05-23
MathArena/aime_2026 Task MathArena/aime_2026Metric MathArena/aime_2026Setup No toolsComparison conditions not established 77.5 Model Card
Reported by a third party		 Evaluated revision not stated 2026-05-23
TIGER-Lab/MMLU-Pro Task mmlu_proMetric mmlu_proComparison conditions not established 77.2 Model Card
Reported by a third party		 Evaluated revision not stated 2026-05-23
cais/hle Task hleMetric hleSetup No toolsComparison conditions not established 5.2 Model Card
Reported by a third party		 Evaluated revision not stated 2026-05-23
llamaindex/ExtractBench Task longMetric longSetup Pipeline name: gemma4_12b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 13.16 ExtractBench
Reported by a third party		 Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task meanMetric meanSetup Pipeline name: gemma4_12b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 45.39 ExtractBench
Reported by a third party		 Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task mediumMetric mediumSetup Pipeline name: gemma4_12b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 30.82 ExtractBench
Reported by a third party		 Evaluated revision not stated 2026-08-24
llamaindex/ExtractBench Task shortMetric shortSetup Pipeline name: gemma4_12b_vllm_extract_oneshot_structured_output_fileComparison conditions not established 53.62 ExtractBench
Reported by a third party		 Evaluated revision not stated 2026-08-24

Memory Requirements

PrecisionWeights in memory
As published23.9 GB
16-bit23.9 GB
8-bit12.0 GB
4-bit6.0 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Built on This Model

Compare gemma-4-12B-it

Questions About gemma-4-12B-it

How much GPU memory does gemma-4-12B-it need?

About 28.7 GB at 16-bit and 7.2 GB at 4-bit: the weights (12B parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run gemma-4-12B-it on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use gemma-4-12B-it commercially?

Yes. gemma-4-12B-it is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

What is gemma-4-12B-it's context length?

262,144 tokens, from the maximum position embeddings in its published configuration.

Similar Models

LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 4-bit quantized version of gemma-4-12B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful…

Open weights apache-2.0 12B parameters 262,144 tokens transformers
View model

LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 8-bit quantized version of gemma-4-12B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful…

Open weights apache-2.0 12B parameters 262,144 tokens transformers
View model

LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 6-bit quantized version of gemma-4-12B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful…

Open weights apache-2.0 12B parameters 262,144 tokens transformers
View model

LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 5-bit quantized version of gemma-4-12B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful…

Open weights apache-2.0 12B parameters 262,144 tokens transformers
View model

Model · Any to any

gemma-4-12B

Google

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages. Featuring both Dense and Mixture-of-Experts (MoE) architectures, Gemma 4 is well-suited for tasks like text generation, coding, and reasoning. The models are available in five distinct sizes: E2B, E4B, 12B, 26B A4B, and 31B. Their diverse sizes make them deployable in environments ranging from…

Open weights apache-2.0 12B parameters 262,144 tokens transformers
View model

Model · Any to any

gemma-4-12B-it-FP8-dynamic

Thor Lin

Self-quantized FP8 (dynamic) of google/gemma-4-12B-it — Google's encoder-free omni model (text + image + audio + video). Quantized and benchmarked on an NVIDIA DGX Spark (GB10, sm121a). TL;DR: 13 GB on disk (from 23 GB BF16), 15.9 tok/s on a GB10 via vLLM, all four modalities intact. Data-free — no calibration needed. If you want the smallest + fastest build, see the sibling NVFP4 weight-only repo. FP8 is the conservative choice (dynamic activations, no calibration, widest kernel support). I scored all three formats on MMLU (English, 57 subjects) and TMMLU+ (Traditional Chinese, 66 subjects) with lm-evaluation-harness, 5-shot, chat template applied, limit=30 (N ≈ 1,710 EN / 1,980 TC, ±~1.0…

Open weights apache-2.0 12B parameters 131,072 tokens transformers
View model