Oossa

Saluki 27B: 2‑bit Qwen3.8 model outperforms original on tool calling

The open‑source Saluki 27B model, only 7.89 GB, beats the full‑size Qwen3.8‑27B at tool‑calling tasks.

NoteBy Published by Oossa: 1 min read

The community‑released Saluki 27B is a 2‑bit GGUF version of Qwen3.8‑27B. It is 7.89 GB in size and licensed under Apache 2.0. According to the MarkTechPost article, Saluki beats the original 54 GB model on tool‑calling benchmarks, though it lags behind on competition‑math and reasoning tests.

Why it matters

Developers can run a much smaller, freely licensed model that handles tool‑calling more effectively on modest hardware.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.