# ModelCloud/GPTQModel

LLM model quantization (compression) toolkit with HW acceleration support for Nvidia, AMD, Intel GPU and Intel/AMD/Apple CPU via HF, vLLM, and SGLang.

Repository: https://github.com/ModelCloud/GPTQModel
Canonical: https://ross.abutalabs.com/products/gptqmodel
Homepage: https://x.com/Qubitium
Language: Python
License: NOASSERTION
License Family: other
Topics: gptq, optimum, peft, quantization, sglang, transformers, vllm
Last push: 2026-09-03T02:02:48+00:00

## Health v2 (maintenance only)
Score: 91/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 100, release rhythm 100, longevity 57
- inputs: {"age_days": 807, "days_push": 0, "days_rel": 2, "gap_med": 6.0, "n_releases_24m": 57}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1248, forks 205 (observed 2026-09-03T02:15:06.704891+00:00)

## What it is
GPTQModel is a production-ready Python toolkit for quantizing (compressing) large language models using GPTQ, AWQ, and related methods, with hardware-accelerated inference across NVIDIA, AMD, Intel, Huawei, and Apple hardware. It integrates with Hugging Face Transformers, vLLM, and SGLang for loading and serving quantized models.

## Use cases
- quantize an LLM to 4-bit with GPTQ
- convert a Hugging Face model to AWQ format
- run quantized models on vLLM or SGLang
- compress large language models for consumer GPUs
- quantize models on AMD ROCm or Apple Silicon
- quantize MoE and multimodal models like DeepSeek or Qwen-VL

## When to choose
- you need to quantize LLMs with GPTQ/AWQ and broad model architecture support
- you want hardware-accelerated inference across multiple GPU vendors
- you need integration with HF Transformers, vLLM, and SGLang

## When to avoid
- you need quantization of non-LLM models like CNNs
- you want a GUI-based one-click quantization tool
- you only need to run pre-quantized models without quantizing your own

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, llm-inference, llm-training, gpu-computing
- domain: large-language-models, machine-learning, deep-learning, gpu-computing
- platform: python, windows
- tags: quantization, gptq, awq, vllm, sglang, huggingface, model-compression, inference-optimization, linux, macos, gpu, cuda

## Member repositories
- ModelCloud/GPTQModel (main) score 91

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:06.704891+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:08:01.117342+00:00, confidence not recorded.
  - readme: https://github.com/ModelCloud/GPTQModel (fetched 2026-09-03T02:15:06.704891+00:00, sha 3e33064ed882)
  - registry_pypi: https://pypi.org/pypi/gptqmodel/json (fetched 2026-08-29T12:19:19.963942+00:00, sha 55c74c0a70ef)
- Data as of 2026-08-30T08:39:29.467469+00:00.