Bartowski releases Xing4.0-29B-A4B quantized models for llama.cpp
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…
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**
The story so far
2 episodes →- Bartowski releases Xing4.0-29B-A4B quantized models for llama.cppthis story
bartowski/XingChen-AGI_Xing4.0-29B-A4B-GGUF · Hugging Face
huggingface.co · 6 November 2026
Loading the full article…
This text was published by huggingface.co. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- MIT CSAIL: GPT-6 Astra edits Qwen's parameters · 1 src
- PSI proves Aaronson-Ambainis conjecture using Anthropic’s non-public model · 1 src
- Mit develops CheMeleon model that outperforms other methods · 1 src
- OpenAI releases BioBasicsBench benchmark; GPT‑6 Astra scores 66.7% on wet‑lab tasks · 1 src
- Baseten releases Serverless Loops for open-model fine-tuning with per-token billing · 1 src
Comments
via GitHub Discussions