What are the hardware requirements for Kiqu 70B AWQ?

Kiqu 70B AWQ requires approximately 36.7 GB of VRAM and supports a context window of 31K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Kiqu 70B AWQ and how large is it?

Kiqu 70B AWQ is developed by maywell, a model with 70b 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 Kiqu 70B AWQ?

Kiqu 70B AWQ is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Kiqu 70B AWQ by maywell — VRAM 36.7GB, 31K context

Name: Kiqu 70B AWQ
Author: maywell

Kiqu 70B AWQ is an open-source language model by maywell. Features: 70b LLM, VRAM: 36.7GB, Context: 31K, License: cc-by-sa-4.0, Quantized, LLM Explorer Score: 0.11.

4-bit Awq Conversational En Endpoints compatible Ko Llama Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/maywell/kiqu-70b-awq

Kiqu 70B AWQ Parameters and Internals

Model Type

text generation

Use Cases

Areas:

research, education

Applications:

chatbots, language translation

Primary Use Cases:

Korean language processing, conversion for chat applications

Considerations:

Few-shot usage is highly recommended and trailing spaces in instruction format should be avoided.

Additional Notes

This is an early leaked version. Commercial usage is risky.

Supported Languages

Korean (high), English (medium)

Training Details

Data Sources:

Korean datasets

Methodology:

SFT+DPO training based on Miqu-70B-Alpaca-DPO

Model Architecture:

Follows Mistral format

Input Output

Input Format:

[INST] {instruction} [/INST] {output}

Accepted Modalities:

text

Output Format:

text responses

Performance Tips:

Avoid trailing spaces after [/INST] to maintain performance.

LLM Name	Kiqu 70B AWQ
Repository 🤗	https://huggingface.co/maywell/kiqu-70b-awq
Model Size	70b
Required VRAM	36.7 GB
Updated	2026-07-28
Maintainer	maywell
Model Type	llama
Model Files	5.0 GB: 1-of-8 5.0 GB: 2-of-8 4.9 GB: 3-of-8 4.9 GB: 4-of-8 4.9 GB: 5-of-8 4.9 GB: 6-of-8 4.9 GB: 7-of-8 2.2 GB: 8-of-8
Supported Languages	ko en
AWQ Quantization	Yes
Quantization Type	awq
Model Architecture	LlamaForCausalLM
License	cc-by-sa-4.0
Context Length	32764
Model Max Length	32764
Transformers Version	4.37.2
Tokenizer Class	LlamaTokenizer
Padding Token	<unk>
Vocabulary Size	32000
Torch Data Type	float16

Rank the Kiqu 70B AWQ 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 Kiqu 70B AWQ

Best Alternatives	Context / RAM	Downloads	Likes
...0B Instruct Gradient 1048K AWQ	1024K / 39.9 GB	5	1
...70B Instruct Gradient 262K AWQ	256K / 39.9 GB	5	0
Llama 3.3 70B Instruct AWQ	128K / 39.9 GB	262002	46
...lama 3.3 70B Instruct AWQ INT4	128K / 39.9 GB	82744	30
Llama 3.3 70B Instruct AWQ	128K / 39.9 GB	16090	12
... SauerkrautLM 70B Instruct AWQ	128K / 39.9 GB	5	5
MultiVerse 70B AWQ	32K / 41.3 GB	70	2
Opus V1.2 70B AWQ	32K / 36.7 GB	9	1
QuartetAnemoi 70B T0.0001 AWQ	31K / 36.7 GB	3	1
Senku 70B AWQ 4bit GEMM	31K / 36.7 GB	2	1

Note: green Score (e.g. "73.2") means that the model is better than maywell/kiqu-70b-awq.