Skip to page content
SAVRN
Search Plan a project

Open-weight model

gemma-4-12B-it-heretic

by Sh0ck0r sh0ck0r/gemma-4-12B-it-heretic

gemma-4-12B-it-heretic is an open-weight model from Sh0ck0r. It has 12B parameters and a 262,144-token context. At 16-bit it needs about 28.7 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 69 downloads a month.

Parameters12B
Context262,144
Weights47.8 GB
License—
AccessOpen weights
Monthly Downloads69

Runs On

What it takes to serve gemma-4-12B-it-heretic (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 Oct 18, 2026.

gemma-4-12B-it-heretic on every accelerator the SAVRN Index prices, at every precision

Model Card

The publisher has not written a card for this model.

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
sh0ck0r/gemma-4-12B-it-heretic
Publisher
Sh0ck0r
Task
Not stated by the source
Modality
Other
Library
Not stated by the source
Parameters
12B parameters
Languages
Not stated by the source
Revision
108d1ece370659a31a61f73cad892711e7e2a480
First published
2026-08-05
Last updated
2026-10-10

Files and Weights

48 files, 71.9 GB in total. The weights are 10 files totalling 47.8 GB in safetensors.

Weights10 files · 47.8 GB
Configuration15 files · 147.3 KB
Tokenizer5 files · 64.3 MB
Other16 files · 24.0 GB
Repository2 files · 3.8 KB
Every file
FileTypeSizeSHA-256
model-00001-of-00005.safetensorsWeights5.0 GB a813e9d92a15
model-00002-of-00005.safetensorsWeights5.0 GB ca356ee69c7e
model-00003-of-00005.safetensorsWeights4.9 GB ebf3d51f05aa
model-00004-of-00005.safetensorsWeights5.0 GB 360b3ba8bfd1
model-00005-of-00005.safetensorsWeights4.1 GB ec82ae7f32d4
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model-00001-of-00005.safetensorsWeights5.0 GB a813e9d92a15
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model-00002-of-00005.safetensorsWeights5.0 GB ca356ee69c7e
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model-00003-of-00005.safetensorsWeights4.9 GB ebf3d51f05aa
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model-00004-of-00005.safetensorsWeights5.0 GB 360b3ba8bfd1
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model-00005-of-00005.safetensorsWeights4.1 GB ec82ae7f32d4
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/added_tokens.jsonConfiguration —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/chat_template.jsonConfiguration —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model_index.jsonConfiguration —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/preprocessor_config.jsonConfiguration —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/special_tokens_map.jsonConfiguration —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/video_preprocessor_config.jsonConfiguration —
config.jsonConfiguration4.3 KB —
generation_config.jsonConfiguration260 B —
model.safetensors.index.jsonConfiguration66.3 KB —
processor_config.jsonConfiguration1.4 KB —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/config.jsonConfiguration4.3 KB —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/generation_config.jsonConfiguration260 B —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/model.safetensors.index.jsonConfiguration66.3 KB —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/processor_config.jsonConfiguration1.4 KB —
trees/55be9808272ae15883b9ccaee1e7ea651c8bdf9a.jsonConfiguration2.7 KB —
blobs/360b3ba8bfd1a38ee0cb33c78b4547edb7092612b63516d083f1264dc6d78650Other5.0 GB 360b3ba8bfd1
blobs/52373fe24473b1aa44333d318f578ae6bf04b49bOther1.6 KB —
blobs/752efb73998ce28a90eb0a1ce0e9a1891e55ec79Other66.3 KB —
blobs/a813e9d92a15d0958c29a9f3c98f1bf8a52b262d3234cbb06886b37c81e1fb28Other5.0 GB a813e9d92a15
blobs/b889adcd78015bf5a0139438c5fb0b7067529a61Other1.4 KB —
blobs/ca356ee69c7ed3d6670abb9d2f31671d3999d761eb0f7ec81621cb21806be802Other5.0 GB ca356ee69c7e
blobs/ca875cd6e42132b385f9140cb2f1a99eab79da1fOther4.3 KB —
blobs/cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0fOther32.2 MB cc8d3a0ce364
blobs/d09dccf1b9aee197ef72a6e88a1ceff74b842ff3Other260 B —
blobs/df4afd629c3e51c2a7ef4e72b967a3f17acbffe1Other2.7 KB —
blobs/e61bbfe9a31db79ea1e4e557b046f895ad7899d0Other17.5 KB —
blobs/ebf3d51f05aac402756bf4733a40fdf0b3b629c670c0f085bc0c195febefa6f1Other4.9 GB ebf3d51f05aa
blobs/ec82ae7f32d4e898d6bf8834f62b30a5b185c71d296830871399acb55bc80531Other4.1 GB ec82ae7f32d4
chat_template.jinjaOther17.5 KB —
refs/mainOther40 B —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/chat_template.jinjaOther17.5 KB —
.gitattributesRepository2.2 KB —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/.gitattributesRepository1.6 KB —
.no_exist/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/audio_tokenizer_config.jsonTokenizer —
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/tokenizer.jsonTokenizer32.2 MB cc8d3a0ce364
snapshots/55be9808272ae15883b9ccaee1e7ea651c8bdf9a/tokenizer_config.jsonTokenizer2.7 KB —
tokenizer.jsonTokenizer32.2 MB cc8d3a0ce364
tokenizer_config.jsonTokenizer2.7 KB —

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download size
47.8 GB
Download from Sh0ck0r

Released by Sh0ck0r through its official repository on Hugging Face.

Memory Requirements

PrecisionWeights in memory
As published47.8 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.

Questions About gemma-4-12B-it-heretic

How much GPU memory does gemma-4-12B-it-heretic 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-heretic 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.

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

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