# Wed Oct 28 02:45:44 PM +03 2026 mode=single server=127.0.0.1:18020
ROW cohort C1 real prompts T=default | e2e=186.55 tok/s | decode(C/meanTPOT)=190.1 | tok/step=2.79 | meanTTFT=82.74 ms
ROW cohort C2 real prompts T=default | e2e=340.00 tok/s | decode(C/meanTPOT)=387.6 | tok/step=2.72 | meanTTFT=110.92 ms
ROW cohort C4 real prompts T=default | e2e=596.96 tok/s | decode(C/meanTPOT)=664.5 | tok/step=2.78 | meanTTFT=145.42 ms
ROW cohort C8 real prompts T=default | e2e=463.88 tok/s | decode(C/meanTPOT)=545.3 | tok/step=2.71 | meanTTFT=366.49 ms
ROW cohort C1 real prompts T=0 | e2e=209.43 tok/s | decode(C/meanTPOT)=214.6 | tok/step=2.85 | meanTTFT=76.98 ms
ROW cohort C2 real prompts T=0 | e2e=353.25 tok/s | decode(C/meanTPOT)=390.6 | tok/step=2.74 | meanTTFT=111.48 ms
ROW cohort C4 real prompts T=0 | e2e=573.89 tok/s | decode(C/meanTPOT)=684.9 | tok/step=2.80 | meanTTFT=129.16 ms
ROW cohort C8 real prompts T=0 | e2e=485.06 tok/s | decode(C/meanTPOT)=575.5 | tok/step=2.85 | meanTTFT=325.05 ms
# raw logs in /home/zeka/ai/benchmarks/hyperqwen-5090-2026-10-28/results/single/run1
