What are the hardware requirements for Llama 30B Instruct?

Llama 30B Instruct requires approximately 65.2 GB of VRAM and supports a context window of 2K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Llama 30B Instruct and how large is it?

Llama 30B Instruct is developed by upstage, a model with 30b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Llama 30B Instruct perform on standard benchmarks?

Llama 30B Instruct has the following published scores: MMLU 59.37. Compare against reference models on this page or on the LLM Explorer leaderboards.

Llama 30B Instruct by upstage — VRAM 65.2GB, 2K context

Name: Llama 30B Instruct
Author: upstage

Llama 30B Instruct is an open-source language model by upstage. Features: 30b LLM, VRAM: 65.2GB, Context: 2K, Instruction-Based, LLM Explorer Score: 0.11, Arc: 62.5, HellaSwag: 86.2, MMLU: 59.4, GSM8K: 12.1.

Dataset:gair/lima Dataset:metaeval/scienceqa tex... Dataset:open-orca/openorca Dataset:openbookqa Dataset:sciq En Endpoints compatible Instruct Instruction Llama Pytorch Region:us Sharded Upstage

Model Card on HF 🤗: https://huggingface.co/upstage/llama-30b-instruct

Llama 30B Instruct Benchmarks

ARC: 62.46 vs 96.7 (so35)^-35.4%

HellaSwag: 86.23 vs 95.3 (gpt4)^-9.5%

MMLU: 59.37 vs 88.3 (so35)^-32.8%

GSM8K: 12.13 vs 96.4 (so35)^-87.4%

LLME Score: 0.11152

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Llama 30B Instruct (upstage/llama-30b-instruct)

🌟 Advertise your project 🚀

Llama 30B Instruct Parameters and Internals

Model Type

Text Generation

Use Cases

Areas:

Research, Commercial Applications

Additional Notes

Model can handle longer inputs with special configurations

Supported Languages

English (Proficient)

Training Details

Data Sources:

openbookqa, sciq, Open-Orca/OpenOrca, metaeval/ScienceQA_text_only, GAIR/lima

Methodology:

Fine-tuning

Context Length:

1024

Hardware Used:

A100x8 * 1

Model Architecture:

Transformer-based

Input Output

Input Format:

### System: {System} ### User: {User} ### Assistant: {Assistant}

Accepted Modalities:

Text

Output Format:

Text

Performance Tips:

The model can handle up to 10k+ input tokens with `rope_scaling` enabled

LLM Name	Llama 30B Instruct
Repository 🤗	https://huggingface.co/upstage/llama-30b-instruct
Model Size	30b
Required VRAM	65.2 GB
Updated	2026-05-24
Maintainer	upstage
Model Type	llama
Instruction-Based	Yes
Model Files	9.8 GB: 1-of-7 10.0 GB: 2-of-7 9.9 GB: 3-of-7 9.9 GB: 4-of-7 9.9 GB: 5-of-7 10.0 GB: 6-of-7 5.7 GB: 7-of-7
Supported Languages	en
Model Architecture	LlamaForCausalLM
Context Length	2048
Model Max Length	2048
Transformers Version	4.30.2
Tokenizer Class	LlamaTokenizer
Beginning of Sentence Token	<s>
End of Sentence Token	</s>
Unk Token	<unk>
Vocabulary Size	32000
Torch Data Type	bfloat16

Best Alternatives to Llama 30B Instruct

Best Alternatives	Context / RAM	Downloads	Likes
Tildeopen 30B Mu Instruct	64K / 60.9 GB	198	5
...ct 2048 Open Platypus 100steps	2K / 65.2 GB	982	0
Llama 30B Instruct 2048	2K / 65.2 GB	1068	103
...azarus Instruct PL Lora Unload	2K / 65.2 GB	205	0
...B 2048 Instruct PL Lora Unload	2K / 65.2 GB	92	1
...lama 30B Instruct 2048 PL Lora	2K / 65.2 GB	201	0
H2ogpt Oasst1 512 30B HF	2K / 65 GB	1007	2
...pt Research Oig Oasst1 512 30B	2K / 81.1 GB	1008	3
Open Instruct Human Mix 30B	2K / 130.5 GB	10	1
...ruct Heretic Safety Defiltered	0.5K / 0.1 GB	10	2

Note: green Score (e.g. "73.2") means that the model is better than upstage/llama-30b-instruct.

Rank the Llama 30B Instruct Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

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

What open-source LLMs or SLMs are you in search of? 53999 in total.

Email us: info@extractum.io. Our Privacy Policy | Terms and Conditions | Suggest an improvement.

Our Social Media →

Original data from HuggingFace, OpenCompass and various public git repos.

Check out Ag3ntum — our secure, self-hosted AI agent for server management.

Release v20260328a

Support LLM Explorer

Llama 30B Instruct by upstage

» All LLMs » upstage » Llama 30B Instruct URL Share it on

Llama 30B Instruct Benchmarks

Llama 30B Instruct Parameters and Internals

Best Alternatives to Llama 30B Instruct

Rank the Llama 30B Instruct Capabilities

What open-source LLMs or SLMs are you in search of? 53999 in total.