What are the hardware requirements for Caelum G4 38B A12.5B?

Caelum G4 38B A12.5B requires approximately 75.7 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Caelum G4 38B A12.5B and how large is it?

Caelum G4 38B A12.5B is developed by EryriLabs, a model with 12b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Caelum G4 38B A12.5B?

Caelum G4 38B A12.5B is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Caelum G4 38B A12.5B by EryriLabs — VRAM 75.7GB

Name: Caelum G4 38B A12.5B
Author: EryriLabs

Caelum G4 38B A12.5B is an open-source language model by EryriLabs. Features: 12b LLM, VRAM: 75.7GB, License: apache-2.0, LLM Explorer Score: 0.3.

Agentic Base model:google/gemma-4-12b-... Base model:merge:google/gemma-... Base model:merge:valiantlabs/g... Base model:merge:valiantlabs/g... Base model:merge:yuxinlu1/gemm... Base model:valiantlabs/gemma-4... Base model:valiantlabs/gemma-4... Base model:yuxinlu1/gemma-4-12... Caelum gemma4 unified moe Coding Conversational Custom-code Custom code Gemma4 Gemma4-unified Image-text-to-text Mergekit Mixture-of-experts Moe Reasoning Region:us Safetensors Sharded Tensorflow Tool-use Trust-remote-code

Model Card on HF 🤗: https://huggingface.co/EryriLabs/Caelum-G4-38B-A12.5B

Caelum G4 38B A12.5B Parameters and Internals

LLM Name	Caelum G4 38B A12.5B
Repository 🤗	https://huggingface.co/EryriLabs/Caelum-G4-38B-A12.5B
Base Model(s)	Gemma 4 12B It yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2 Gemma 4 12B It Esper4 Gemma 4 12B It Guardpoint google/gemma-4-12B-it yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2 ValiantLabs/gemma-4-12B-it-Esper4 ValiantLabs/gemma-4-12B-it-Guardpoint
Model Size	12b
Required VRAM	75.7 GB
Updated	2026-08-10
Maintainer	EryriLabs
Model Type	caelum_gemma4_unified_moe
Model Files	0.0 GB 5.0 GB: 1-of-18 4.8 GB: 2-of-18 4.3 GB: 3-of-18 4.3 GB: 4-of-18 4.3 GB: 5-of-18 4.3 GB: 6-of-18 4.3 GB: 7-of-18 4.3 GB: 8-of-18 4.3 GB: 9-of-18 4.3 GB: 10-of-18 4.3 GB: 11-of-18 4.3 GB: 12-of-18 4.3 GB: 13-of-18 4.3 GB: 14-of-18 4.3 GB: 15-of-18 4.3 GB: 16-of-18 4.3 GB: 17-of-18 1.4 GB: 18-of-18
Model Architecture	Gemma4UnifiedForConditionalGeneration
License	apache-2.0
Transformers Version	5.14.1
Tokenizer Class	GemmaTokenizer
Padding Token	<pad>

Rank the Caelum G4 38B A12.5B capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Caelum G4 38B A12.5B

Best Alternatives	Context / RAM	Downloads	Likes
Gemma 4 12B It	0K / 23.9 GB	2965396	1417
Gemma 4 12B	0K / 23.9 GB	103595	699
Gemma 4 12B It NVFP4	0K / 9.3 GB	47579	20
Gemma 4 12B It	0K / 23.9 GB	146843	15
...le5 Composer2.5 V1 Abliterated	0K / 24 GB	30003	1
Grug 12B	0K / 23.9 GB	519	67
...2B Coder Fable5 Composer2.5 V1	0K / 23.9 GB	1477	58
Gemma 4 12B	0K / 23.9 GB	13026	18
Gemma 4 12B StyleTune	0K / 25.9 GB	1191	27
Gemma 4 12B It Esper4	0K / 23.9 GB	422	10

Note: green Score (e.g. "73.2") means that the model is better than EryriLabs/Caelum-G4-38B-A12.5B.