What are the hardware requirements for Mistral Ko Inst Dev?

Mistral Ko Inst Dev requires approximately 14.8 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 Mistral Ko Inst Dev and how large is it?

Mistral Ko Inst Dev is developed by beomi, a model with 7.4b 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 Mistral Ko Inst Dev?

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

Mistral Ko Inst Dev by beomi — VRAM 14.8GB, 32K context

Name: Mistral Ko Inst Dev
Author: beomi

Mistral Ko Inst Dev is an open-source language model by beomi. Features: 7.4b LLM, VRAM: 14.8GB, Context: 32K, License: apache-2.0, LLM Explorer Score: 0.11.

Conversational Dataset:heegyu/kowiki-sentence... Dataset:heegyu/kowikitext En Endpoints compatible Ko Mistral Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/beomi/Mistral-Ko-Inst-dev

Mistral Ko Inst Dev Parameters and Internals

Model Type

text generation, natural language understanding

Use Cases

Areas:

Research, Experimental applications

Applications:

Korean language processing, NLU tasks

Primary Use Cases:

Answering in Korean using English knowledge

Limitations:

Designed for experimental use, contents updated without notice, Minimal dataset usage may impact performance and accuracy

Considerations:

Use the repository with a specific git hash for stable results

Additional Notes

This is an experimental repository and content updates may occur without notice.

Supported Languages

ko (proficient), en (proficient)

Training Details

Data Sources:

heegyu/kowikitext, heegyu/kowiki-sentences

Data Volume:

Minimal dataset used: Korean Wikipedia only

Methodology:

Adapt new Korean vocabulary, maintain NLU capability of Mistral-Instruct model

Input Output

Input Format:

Text in Korean or English

Accepted Modalities:

text

Output Format:

Generated text responses in Korean or English

LLM Name	Mistral Ko Inst Dev
Repository 🤗	https://huggingface.co/beomi/Mistral-Ko-Inst-dev
Model Size	7.4b
Required VRAM	14.8 GB
Updated	2026-08-03
Maintainer	beomi
Model Type	mistral
Model Files	5.0 GB: 1-of-3 4.9 GB: 2-of-3 4.9 GB: 3-of-3
Supported Languages	ko en
Model Architecture	MistralForCausalLM
License	apache-2.0
Context Length	32768
Model Max Length	32768
Transformers Version	4.38.2
Tokenizer Class	LlamaTokenizer
Vocabulary Size	46592
Torch Data Type	bfloat16

Rank the Mistral Ko Inst Dev 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 Mistral Ko Inst Dev

Best Alternatives	Context / RAM	Downloads	Likes
Tamtanai	32K / 14.8 GB	10	0
... Phi 3 Medium Bnb 4bit Smashed	4K / 8.4 GB	5	0

Note: green Score (e.g. "73.2") means that the model is better than beomi/Mistral-Ko-Inst-dev.