# Wed Oct 28 02:50:30 PM +03 2026 mode=single server=127.0.0.1:18020
ROW cohort C1 real prompts T=default | e2e=198.60 tok/s | decode(C/meanTPOT)=201.6 | tok/step=2.74 | meanTTFT=79.45 ms
ROW cohort C2 real prompts T=default | e2e=364.92 tok/s | decode(C/meanTPOT)=389.9 | tok/step=2.77 | meanTTFT=109.39 ms
ROW cohort C4 real prompts T=default | e2e=557.07 tok/s | decode(C/meanTPOT)=665.6 | tok/step=2.79 | meanTTFT=135.05 ms
ROW cohort C8 real prompts T=default | e2e=473.40 tok/s | decode(C/meanTPOT)=552.5 | tok/step=2.72 | meanTTFT=334.92 ms
ROW cohort C1 real prompts T=0 | e2e=208.01 tok/s | decode(C/meanTPOT)=212.8 | tok/step=2.85 | meanTTFT=77.61 ms
ROW cohort C2 real prompts T=0 | e2e=336.58 tok/s | decode(C/meanTPOT)=372.4 | tok/step=2.74 | meanTTFT=114.86 ms
ROW cohort C4 real prompts T=0 | e2e=550.41 tok/s | decode(C/meanTPOT)=656.8 | tok/step=2.80 | meanTTFT=136.99 ms
ROW cohort C8 real prompts T=0 | e2e=475.97 tok/s | decode(C/meanTPOT)=569.0 | tok/step=2.85 | meanTTFT=363.66 ms
# raw logs in /home/zeka/ai/benchmarks/hyperqwen-5090-2026-10-28/results/single/run2
