1M Tokens/s: Scaling Qwen 3.5 27B on 96 B200 GPUs with vLLM

(medium.com) by m4r1k | Mar 27, 2026 | 0 comments on HN