What are the hardware requirements for Underground?

Underground requires approximately 21.4 GB of VRAM and supports a context window of 4K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Underground and how large is it?

Underground is developed by ClaudioItaly, a model with 11b 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 Underground?

Underground is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Underground by ClaudioItaly — VRAM 21.4GB, 4K context

Name: Underground
Author: ClaudioItaly

Underground is an open-source language model by ClaudioItaly. Features: 11b LLM, VRAM: 21.4GB, Context: 4K, Merged, LLM Explorer Score: 0.13.

Merged Model Base model:bhavinjawade/solar-... Base model:sao10k/fimbulvetr-1... Endpoints compatible Llama Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/ClaudioItaly/Underground

Underground Parameters and Internals

Additional Notes

This model is a merging of pre-trained language models created using mergekit with the SLERP merge method. The merged models are Sao10K/Fimbulvetr-11B-v2 and bhavinjawade/SOLAR-10B-Nector-DPO-Jawade.

LLM Name	Underground
Repository 🤗	https://huggingface.co/ClaudioItaly/Underground
Base Model(s)	Fimbulvetr 11B V2 SOLAR 10B Nector DPO Jawade Sao10K/Fimbulvetr-11B-v2 bhavinjawade/SOLAR-10B-Nector-DPO-Jawade
Merged Model	Yes
Model Size	11b
Required VRAM	21.4 GB
Updated	2026-08-25
Maintainer	ClaudioItaly
Model Type	llama
Model Files	4.9 GB: 1-of-5 4.9 GB: 2-of-5 4.9 GB: 3-of-5 5.0 GB: 4-of-5 1.7 GB: 5-of-5
Model Architecture	LlamaForCausalLM
Context Length	4096
Model Max Length	4096
Transformers Version	4.41.0
Tokenizer Class	LlamaTokenizer
Vocabulary Size	32000
Torch Data Type	bfloat16

Rank the Underground 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 Underground

Best Alternatives	Context / RAM	Downloads	Likes
A6.1	1024K / 16.1 GB	130	0
...k PL 11B V3.0 Instruct Heretic	32K / 22.3 GB	149	0
MIstral 11B Omni OP U1k Ver0.1	32K / 21.4 GB	11	0
...ral 11B Omni OP 1K 2048 Ver0.1	32K / 21.4 GB	8	0
Llama 3 Synatra 11B V1 20K	20K / 23 GB	4	9
Fimbulvetr 11B V2.1 16K	16K / 21.4 GB	39	17
Moistral 11B V2	8K / 21.4 GB	59	21
Narumashi 11B V0.9	8K / 21.4 GB	54	1
Moistral 11B V5d E4	8K / 21.4 GB	15	1
Moistral 11B V3	8K / 21.4 GB	67	108

Note: green Score (e.g. "73.2") means that the model is better than ClaudioItaly/Underground.