# Bartowski releases Xing4.0-29B-A4B quantized models for llama.cpp

Digest AI · Research · published 2026-11-06T19:19:24Z

Canonical: https://digestai.news/story/bartowski-releases-xing4-0-29b-a4b-quantized-models-for-llama-cpp

## Summary

A user named bartowski has shared quantized versions of the **Xing4.0-29B-A4B** model on Hugging Face, optimized for **llama.cpp** with **speculative decoding** and **imatrix calibration**. The original model has **31 billion parameters** and supports text input. Quantized versions range from **20.06 GB (Q4\(K\)M)** to smaller sizes, designed for faster inference on CPUs and GPUs. The quantizations include custom layouts tuned for performance, with details on trade-offs between speed and quality, such as **Q4\(K\)M** (90% base precision) and **IQ3\(M\)** (GPU-optimized). The files are compatible with tools like **LM Studio, Jan AI, and Text Generation Web UI**, and the author provides calibration data and instructions for running them efficiently on hardware with limited RAM or VRAM.

## Key points

- bartowski released quantized versions of **Xing4.0-29B-A4B** for **llama.cpp**, with **speculative decoding** and **imatrix calibration**
- Largest quantized file is **20.06 GB (Q4\(K\)M)**, optimized for speed and quality trade-offs on CPUs/GPUs
- Models support **text input only** and work with tools like **LM Studio, Jan AI, and Text Generation Web UI**

## Why it matters

Quantized versions of large models like this enable developers and hobbyists to run advanced AI locally without high-end hardware. The speculative decoding and imatrix optimizations could improve inference speed for text generation tasks, making powerful models more accessible for research and small-scale applications.

## Sources

1. [bartowski/XingChen-AGI_Xing4.0-29B-A4B-GGUF · Hugging Face](https://huggingface.co/bartowski/XingChen-AGI_Xing4.0-29B-A4B-GGUF) (huggingface.co, 2026-11-06, primary source)

Part of the developing story: [Bartowski releases Xing4.0-29B-A4B quantized models for llama.cpp](https://digestai.news/thread/marktechpost-publishes-tutorial-on-running-underdog-saluki-27b-quantized-model) (2 stories)

## Cite

Digest AI, "Bartowski releases Xing4.0-29B-A4B quantized models for llama.cpp", 6 November 2026, https://digestai.news/story/bartowski-releases-xing4-0-29b-a4b-quantized-models-for-llama-cpp

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/bartowski-releases-xing4-0-29b-a4b-quantized-models-for-llama-cpp.json