Saluki 27B claims 96% of Qwen 3.8 performance at 1/7 size

A LocalLLaMA user posted about the Saluki 27B model, claiming it reaches about 96% of Qwen 3.8's performance at roughly one-seventh the size, and asked the community for feedback.

• 4x AMD R9700 local LLM rig for 10-person startup costs $18K2026-10-11
• User says uncensored models reduce work headaches2026-09-24
• Did Alibaba abandon its 35B A3B model?2026-09-22
• Qwen 3.8 27B ran 63 hours on an RTX 30902026-09-16
• Set Up OpenCode with Ollama and sbx on Mac2026-09-11