Edit: Title says 134 tps, it's actually 138 -- keep in mind my 3090 is power limited to 250w. Three days ago I released a hyper-optimized Qwen3.8-27B inference engine for an RTX 3090 (82 tps single request, 672 peak), and yesterday's update took it to ~114 tps single-user / ~1,000 tps at 64 concurre
