Fastest & Best quality & Largest context
Qwen3.8-27B-GPTQ-W4A16
27B
47.7 tok/sQuality 81.8Context 65,536 tokens
Highest token generation speed at 47.7 tok/s. Highest completed quality score at 81.8. Largest context that fits in 23GB VRAM, at 65,536 tokens.