What are the hardware requirements for Quark 270M Instruct?

Quark 270M Instruct requires approximately 0.6 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Quark 270M Instruct and how large is it?

Quark 270M Instruct is developed by ThingAI, a model with 270m 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 Quark 270M Instruct?

Quark 270M Instruct is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Quark 270M Instruct by ThingAI — VRAM 0.6GB

Name: Quark 270M Instruct
Author: ThingAI

Quark 270M Instruct is an open-source language model by ThingAI. Features: 270m LLM, VRAM: 0.6GB, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.25.

Bilingual Chat Conversational Custom code En English Instruct It Italian Quark Region:us Safetensors Sft Small-language-model Trained-from-scratch

Model Card on HF 🤗: https://huggingface.co/ThingAI/Quark-270m-Instruct

Quark 270M Instruct Parameters and Internals

LLM Name	Quark 270M Instruct
Repository 🤗	https://huggingface.co/ThingAI/Quark-270m-Instruct
Model Size	270m
Required VRAM	0.6 GB
Updated	2026-07-23
Maintainer	ThingAI
Model Type	quark
Instruction-Based	Yes
Model Files	0.6 GB
Supported Languages	it en
Model Architecture	QuarkForCausalLM
License	apache-2.0
Model Max Length	2048
Tokenizer Class	TokenizersBackend
Padding Token	<pad>
Vocabulary Size	65537
Torch Data Type	bfloat16

Rank the Quark 270M Instruct 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