Image55 multi turn dialogue latency

Test Date Total Test Cases Concurrency Prefix Cache Enabled Chunk Acceleration Enabled RAG Chunk Length RAG Chunk Recall Count RAG Chunk Hit Count TopK Sparsification Enabled Total Request Time (ms) E2E Throughput (tokens/s) Average Throughput per Request (tokens/s) First Token Latency P50 (ms) First Token Latency P90 (ms) First Token Latency P99 (ms) Max First Token Latency (ms) Average First Token Latency (ms) Incremental Latency P50 (ms) Incremental Latency P90 (ms) Incremental Latency P99 (ms) Max Incremental Latency (ms) Average Incremental Latency (ms) Average Accuracy
2025-10-16-14-31-31 279 1 FALSE FALSE 3 1000 0 FALSE 465072.86 4.9 6.96 1151.86 2930.78 3429.74 3545.17 1452.4 41.47 69.7 105.57 11209.48 56.1 0.45091
2025-10-16-14-31-31 279 1 TRUE FALSE 3 1000 0 FALSE 155829.15 14.63 17.13 217.37 686.87 995.54 1149.71 353.46 32.15 40.1 46.24 58.37 30.34 0.44326
2025-10-16-14-31-31 279 1 TRUE FALSE 3 1000 0 TRUE 158653.1 14.36 16.79 218.61 700.11 981.06 1159.31 356.49 33.18 41.13 47.96 79.15 31.24 0.43334
2025-10-16-14-31-31 279 1 TRUE TRUE 3 1000 0 FALSE 368143.72 6.19 6.71 1307.98 1570.18 1825.02 2107.98 1164.96 42.03 61.49 75.46 116.13 40.64 0.47634
2025-10-16-14-31-31 279 1 TRUE TRUE 3 1000 3 FALSE 366629.04 6.22 6.74 1285.11 1581.82 1864.77 1898.59 1149.23 42.49 63.35 78.72 237.31 41.62 0.37617
2025-10-16-14-31-31 279 1 TRUE TRUE 3 1000 0 TRUE 373218.55 6.11 6.62 1314.76 1601.37 1834.68 2127.12 1167.98 43.4 64.82 79.64 140.89 42.62 0.44979

Comments