vllm serve /workspace/prisma/modelos/Qwen3.8-27B-FP8 --host 127.0.0.1 --port 8000 --served-model-name qwen3.8-27b --tensor-parallel-size 1 --max-model-len 65536 --gpu-memory-utilization 0.92 --enable-auto-tool-choice --tool-call-parser qwen3_xml --reasoning-parser qwen3 --enable-log-requests --enable-log-outputs