What are the hardware requirements for ALIA 40B?

ALIA 40B requires approximately 80.4 GB of VRAM and supports a context window of 32K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed ALIA 40B and how large is it?

ALIA 40B is developed by BSC-LT, a model with 40b 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 ALIA 40B?

ALIA 40B is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

ALIA 40B by BSC-LT — VRAM 80.4GB, 32K context

Name: ALIA 40B
Author: BSC-LT

ALIA 40B is an open-source language model by BSC-LT. Features: 40b LLM, VRAM: 80.4GB, Context: 32K, License: apache-2.0, LLM Explorer Score: 0.17.

Arxiv:1803.09010 Arxiv:1810.06694 Arxiv:1906.03741 Arxiv:1911.05507 Arxiv:2101.00027 Arxiv:2207.00220 Arxiv:2402.06619 Arxiv:2403.14009 Arxiv:2403.20266 Arxiv:2406.17557 Arxiv:2502.08489 Bg Ca Code Cs Cy Da Dataset:ai-team-uoa/greek lega... Dataset:allenai/pes2o Dataset:bigcode/starcoderdata Dataset:bjoernp/tagesschau-201... Dataset:community-datasets/hrw... Dataset:danish-foundation-mode... Dataset:eleutherai/the pile de... Dataset:hitz/euscrawl Dataset:hitz/latxa-corpus-v1.1 Dataset:hoskinson-center/proof... Dataset:huggingfacefw/fineweb-... Dataset:joelniklaus/eurlex res... Dataset:joelniklaus/legal-mc4 Dataset:oscar-corpus/colossal-... Dataset:pile-of-law/pile-of-la... Dataset:pleias/french-pd-books Dataset:pleias/french-pd-newsp... Dataset:portulan/parlamento-pt Dataset:projecte-aina/catalog Dataset:togethercomputer/redpa... Dataset:ufrgs/brwac De Deploy:azure El En Endpoints compatible Es Et Eu Fi Fr Ga Gl Hr Hu It Llama Lt Lv Mt Nl Nn Oc Pl Pt Region:eu Ro Ru Safetensors Sh Sharded Sk Sl Sr Sv Tensorflow Uk

Model Card on HF 🤗: https://huggingface.co/BSC-LT/ALIA-40b

ALIA 40B Parameters and Internals

LLM Name	ALIA 40B
Repository 🤗	https://huggingface.co/BSC-LT/ALIA-40b
Model Size	40b
Required VRAM	80.4 GB
Updated	2026-08-25
Maintainer	BSC-LT
Model Type	llama
Model Files	4.9 GB: 1-of-17 4.9 GB: 2-of-17 4.9 GB: 3-of-17 4.8 GB: 4-of-17 4.9 GB: 5-of-17 4.9 GB: 6-of-17 4.9 GB: 7-of-17 4.8 GB: 8-of-17 4.9 GB: 9-of-17 4.9 GB: 10-of-17 4.9 GB: 11-of-17 4.8 GB: 12-of-17 4.9 GB: 13-of-17 4.9 GB: 14-of-17 4.9 GB: 15-of-17 3.0 GB: 16-of-17 4.2 GB: 17-of-17
Supported Languages	bg ca code cs cy da de el en es et eu fi fr ga gl hr hu it lt lv mt nl nn oc pl pt ro ru sh sk sl sr sv uk
Model Architecture	LlamaForCausalLM
License	apache-2.0
Context Length	32768
Model Max Length	32768
Transformers Version	4.45.2
Tokenizer Class	LlamaTokenizer
Vocabulary Size	256000
Torch Data Type	bfloat16

Rank the ALIA 40B 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 ALIA 40B

Best Alternatives	Context / RAM	Downloads	Likes
ALIA 40B Instruct 2605	160K / 80.4 GB	1109	4
ALIA 40B Instruct 2601	160K / 80.4 GB	1127	12
ALIA 40B Fc 2605	160K / 80.4 GB	79	0
Llama3AnFeng 40B	128K / 80.4 GB	12	0

Note: green Score (e.g. "73.2") means that the model is better than BSC-LT/ALIA-40b.