IT Home On October 10, the artificial intelligence laboratory Underdog AI released the Saluki 27B model this month. This model is a 2-bit quantization optimized version of Qwen3.8-27B, retaining only text I/O capabilities, and has a file size of only 7.89 GB.

Undervalued indicates that Saluki is designed for agent applications. It excels in everyday reasoning and tool usage, while its mathematical performance is relatively weaker. Its tool selection and multi-tool parallel invocation capabilities are even better than those of the full-size Qwen3.8-27B.
The Saluki model uses the standard llama.cpp format, which can be loaded in standard llama.cpp and applications built based on it, without any custom compilation required.
Reference
Ad statement: The text contains external link references (including but not limited to hyperlinks, QR codes, passwords, etc.) designed to provide more information, save selection time, and the results are for reference only. Articles on IT Home that include external links all include this statement.
