Community benchmarks Qwen 3.6 27B with VLLM across quants

Reddit user shares VLLM performance benchmarks for Qwen 3.6 27B in BF16, FP8, and NVFP4 formats. NVFP4 is fastest but has looping issues in copilot.