November 6, 2025 at 03:39 AM
Dataset: reference (v1.0)
Click a metric to highlight the best run in the table below
Performance across different input/output token combinations and concurrency levels
| Input Tokens | Output Tokens | Concurrency | Output TPS | Input TPS | Energy Cost (kWh/MT) | TTFT Mean | TTFT P95 | E2E P95 | Success Rate |
|---|---|---|---|---|---|---|---|---|---|
Best Run for Output TPS | |||||||||
| 128 | 512 | 512x | 5,012.77★ | 1,367.05 | 0.05 | 1,207.76 | 2,328.83 | 41,556.05 | 100.0% |
| 128 | 128 | 1x | 41.29 | 39.35 | 2.03 | 254.43 | 254.43 | 3,099.30 | 100.0% |
| 128 | 128 | 2x | 84.41 | 85.09 | 1.22 | 65.79 | 77.08 | 2,926.04 | 100.0% |
| 128 | 128 | 4x | 169.93 | 168.60 | 0.56 | 105.32 | 118.49 | 3,006.52 | 100.0% |
| 128 | 128 | 8x | 329.47 | 330.44 | 0.30 | 165.28 | 187.68 | 3,095.06 | 100.0% |
| 128 | 128 | 16x | 621.17 | 625.72 | 0.17 | 238.59 | 282.77 | 3,269.81 | 100.0% |
| 128 | 128 | 32x | 1,091.21 | 1,095.07 | 0.12 | 328.89 | 498.57 | 3,545.87 | 96.9% |
| 128 | 128 | 64x | 1,508.75 | 1,517.87 | 0.07 | 1,804.11 | 1,964.82 | 5,295.49 | 100.0% |
| 128 | 128 | 128x | 1,826.68 | 1,906.53 | 0.06 | 2,185.55 | 4,272.60 | 8,152.60 | 100.0% |
| 128 | 128 | 256x | 3,318.26 | 3,473.66 | 0.04 | 678.16 | 1,809.85 | 7,273.48 | 99.2% |
| 128 | 128 | 512x | 3,268.08 | 3,625.44 | 0.04 | 2,074.72 | 4,791.55 | 13,227.93 | 99.8% |
| 128 | 512 | 1x | 30.89 | 28.12 | 2.51 | 1,341.12 | 1,341.12 | 4,314.42 | 100.0% |
| 128 | 512 | 2x | 48.35 | 21.62 | 3.29 | 67.03 | 80.83 | 11,032.70 | 100.0% |
| 128 | 512 | 4x | 175.19 | 43.46 | 1.01 | 81.71 | 89.46 | 11,683.49 | 100.0% |
| 128 | 512 | 8x | 344.90 | 86.48 | 0.52 | 109.71 | 123.80 | 11,868.61 | 100.0% |
| 128 | 512 | 16x | 613.53 | 168.31 | 0.30 | 133.93 | 148.77 | 12,218.42 | 100.0% |
| 128 | 512 | 32x | 1,226.65 | 328.87 | 0.17 | 158.39 | 181.37 | 12,388.53 | 100.0% |
| 128 | 512 | 64x | 2,000.29 | 589.84 | 0.11 | 150.04 | 202.66 | 13,506.85 | 100.0% |
| 128 | 512 | 128x | 3,478.11 | 953.38 | 0.07 | 150.51 | 219.90 | 15,339.96 | 100.0% |
| 128 | 512 | 256x | 4,304.94 | 1,190.98 | 0.05 | 2,206.10 | 3,406.94 | 26,662.42 | 100.0% |
| 128 | 512 | 1024x | 4,188.17 | 1,170.55 | 0.05 | 14,591.35 | 49,854.84 | 81,431.16 | 99.8% |
| 128 | 1,024 | 1x | 44.29 | 7.23 | 4.37 | 81.16 | 81.16 | 16,843.00 | 100.0% |
| 128 | 1,024 | 2x | 88.01 | 10.74 | 2.35 | 69.86 | 80.94 | 23,262.81 | 100.0% |
| 128 | 1,024 | 4x | 175.65 | 21.78 | 1.17 | 99.87 | 112.27 | 23,310.77 | 100.0% |
| 128 | 1,024 | 8x | 187.75 | 38.25 | 1.03 | 147.92 | 166.17 | 23,433.27 | 87.5% |
| 128 | 2,048 | 1x | 44.55 | 2.65 | 4.98 | 52.84 | 52.84 | 45,968.56 | 100.0% |
| 128 | 2,048 | 2x | 47.92 | 6.84 | 4.26 | 70.60 | 80.25 | 34,855.00 | 100.0% |
| 128 | 2,048 | 4x | 141.46 | 10.87 | 1.55 | 82.25 | 90.07 | 46,727.93 | 100.0% |
| 128 | 2,048 | 8x | 192.73 | 21.83 | 1.11 | 111.96 | 126.94 | 47,039.46 | 100.0% |
| 128 | 2,048 | 16x | 517.36 | 41.53 | 0.45 | 207.18 | 256.94 | 49,628.20 | 100.0% |
| 128 | 2,048 | 32x | 831.11 | 76.42 | 0.28 | 1,872.69 | 1,930.93 | 53,633.40 | 100.0% |
| 128 | 2,048 | 64x | 1,804.46 | 135.56 | 0.15 | 3,166.19 | 3,296.33 | 60,022.86 | 100.0% |
| 128 | 2,048 | 128x | 3,153.17 | 238.80 | 0.10 | 434.66 | 1,015.35 | 67,508.88 | 99.2% |
| 128 | 2,048 | 256x | 4,168.83 | 324.51 | 0.07 | 695.73 | 1,796.82 | 98,546.77 | 99.6% |
| 128 | 2,048 | 512x | 3,275.70 | 253.25 | 0.09 | 1,030.68 | 2,911.88 | 243,219.45 | 97.3% |
| 512 | 128 | 1x | 27.29 | 105.73 | 1.14 | 1,841.03 | 1,841.03 | 4,691.30 | 100.0% |
| 512 | 128 | 2x | 84.54 | 328.60 | 0.44 | 125.29 | 137.81 | 3,024.67 | 100.0% |
| 512 | 128 | 4x | 130.69 | 638.27 | 0.27 | 169.69 | 205.97 | 3,090.62 | 100.0% |
| 512 | 128 | 8x | 309.74 | 1,194.80 | 0.15 | 324.17 | 365.70 | 3,296.80 | 100.0% |
| 512 | 128 | 16x | 513.78 | 1,989.83 | 0.10 | 490.93 | 686.56 | 3,713.16 | 93.8% |
| 512 | 128 | 32x | 871.46 | 3,618.45 | 0.05 | 601.48 | 1,255.48 | 4,322.93 | 100.0% |
| 512 | 128 | 64x | 1,371.96 | 5,337.46 | 0.04 | 992.63 | 2,248.07 | 5,820.83 | 100.0% |
| 512 | 128 | 128x | 1,279.56 | 5,285.29 | 0.04 | 1,552.74 | 3,942.16 | 11,707.64 | 100.0% |
| 512 | 512 | 1x | 44.53 | 43.14 | 2.60 | 51.52 | 51.52 | 11,496.49 | 100.0% |
| 512 | 512 | 2x | 57.64 | 85.72 | 1.54 | 60.62 | 62.80 | 11,201.49 | 100.0% |
| 512 | 512 | 4x | 150.64 | 168.92 | 0.70 | 87.11 | 97.61 | 11,705.67 | 100.0% |
| 512 | 512 | 8x | 281.82 | 331.21 | 0.37 | 111.93 | 123.63 | 11,916.28 | 100.0% |
| 512 | 512 | 16x | 564.11 | 640.83 | 0.20 | 139.41 | 154.47 | 12,336.35 | 100.0% |
| 512 | 512 | 32x | 1,207.08 | 1,250.57 | 0.11 | 195.14 | 216.14 | 12,669.04 | 100.0% |
| 512 | 512 | 64x | 2,027.67 | 2,203.33 | 0.07 | 183.36 | 220.95 | 14,000.39 | 98.4% |
| 512 | 512 | 128x | 3,280.11 | 3,616.82 | 0.04 | 167.74 | 236.54 | 16,387.51 | 100.0% |
| 512 | 512 | 256x | 3,363.61 | 3,595.79 | 0.05 | 2,556.10 | 8,444.38 | 33,661.60 | 99.6% |
| 512 | 512 | 512x | 3,525.39 | 3,738.73 | 0.04 | 4,743.98 | 16,061.81 | 61,761.94 | 99.8% |
| 512 | 512 | 1024x | 3,064.30 | 3,280.21 | 0.05 | 44,466.42 | 103,102.34 | 150,797.04 | 99.7% |
| 512 | 1,024 | 1x | 43.60 | 61.26 | 2.07 | 116.65 | 116.65 | 8,072.31 | 100.0% |
| 512 | 1,024 | 2x | 87.39 | 42.46 | 1.76 | 124.19 | 125.54 | 23,432.59 | 100.0% |
| 512 | 1,024 | 4x | 139.61 | 83.98 | 1.05 | 171.44 | 205.21 | 23,541.60 | 100.0% |
| 512 | 1,024 | 8x | 299.35 | 163.75 | 0.52 | 327.10 | 369.45 | 24,112.35 | 100.0% |
| 512 | 1,024 | 16x | 594.20 | 313.06 | 0.27 | 438.19 | 643.26 | 25,317.18 | 100.0% |
| 512 | 1,024 | 32x | 1,159.26 | 598.82 | 0.15 | 585.89 | 1,233.75 | 26,559.08 | 100.0% |
| 512 | 1,024 | 64x | 1,762.74 | 1,050.77 | 0.11 | 928.06 | 2,268.27 | 30,145.46 | 100.0% |
| 512 | 1,024 | 128x | 2,915.04 | 1,644.47 | 0.07 | 1,440.71 | 4,228.85 | 38,012.32 | 99.2% |
| 512 | 1,024 | 256x | 3,554.22 | 2,001.52 | 0.06 | 2,695.96 | 10,761.65 | 61,577.78 | 99.6% |
| 512 | 1,024 | 512x | 2,870.08 | 1,615.53 | 0.07 | 5,023.90 | 19,014.26 | 147,933.91 | 99.4% |
| 512 | 2,048 | 1x | 44.28 | 10.73 | 4.21 | 118.39 | 118.39 | 46,245.10 | 100.0% |
| 512 | 2,048 | 2x | 87.40 | 21.23 | 2.16 | 106.60 | 140.04 | 46,856.96 | 100.0% |
| 512 | 2,048 | 4x | 132.11 | 42.09 | 1.35 | 180.40 | 216.77 | 46,991.79 | 100.0% |
| 512 | 2,048 | 8x | 275.00 | 82.16 | 0.67 | 300.69 | 371.78 | 48,057.92 | 100.0% |
| 512 | 2,048 | 16x | 511.24 | 156.74 | 0.37 | 419.66 | 612.90 | 50,588.87 | 100.0% |
| 512 | 2,048 | 32x | 1,041.27 | 298.19 | 0.21 | 586.13 | 1,248.96 | 53,440.16 | 100.0% |
| 512 | 2,048 | 64x | 1,750.94 | 503.81 | 0.13 | 1,337.95 | 3,652.75 | 62,089.30 | 98.4% |
| 512 | 2,048 | 128x | 2,768.79 | 870.62 | 0.09 | 1,514.77 | 4,338.41 | 72,688.35 | 100.0% |
| 512 | 2,048 | 256x | 3,396.65 | 1,058.94 | 0.07 | 2,557.09 | 8,492.04 | 114,589.12 | 100.0% |
| 512 | 2,048 | 512x | 2,699.66 | 833.68 | 0.09 | 7,431.07 | 16,795.97 | 254,305.72 | 85.0% |
| 1,024 | 128 | 1x | 40.59 | 465.62 | 0.40 | 185.54 | 185.54 | 2,070.89 | 100.0% |
| 1,024 | 128 | 2x | 43.19 | 644.25 | 0.30 | 183.03 | 184.09 | 2,888.85 | 100.0% |
| 1,024 | 128 | 4x | 121.69 | 1,204.25 | 0.16 | 280.14 | 350.32 | 3,237.91 | 100.0% |
| 1,024 | 128 | 8x | 283.66 | 2,169.81 | 0.10 | 571.54 | 648.56 | 3,599.26 | 100.0% |
| 1,024 | 128 | 16x | 471.95 | 3,641.86 | 0.06 | 713.23 | 1,196.15 | 4,285.64 | 100.0% |
| 1,024 | 128 | 32x | 712.91 | 5,603.96 | 0.04 | 946.23 | 2,329.34 | 5,493.34 | 100.0% |
| 1,024 | 128 | 64x | 922.09 | 7,386.54 | 0.04 | 1,697.68 | 4,372.09 | 8,209.11 | 98.4% |
| 1,024 | 128 | 128x | 1,132.20 | 9,154.83 | 0.03 | 2,622.74 | 8,712.84 | 13,295.58 | 100.0% |
| 1,024 | 128 | 256x | 1,208.82 | 9,723.98 | 0.03 | 4,885.67 | 16,515.44 | 24,465.96 | 100.0% |
| 1,024 | 128 | 512x | 1,214.92 | 9,762.92 | 0.03 | 9,628.29 | 34,026.05 | 45,020.51 | 99.4% |
| 1,024 | 128 | 1024x | 621.78 | 4,996.26 | 0.05 | 58,609.88 | 115,659.72 | 131,302.35 | 66.8% |
| 1,024 | 512 | 1x | 43.92 | 83.63 | 1.75 | 181.28 | 181.28 | 11,657.09 | 100.0% |
| 1,024 | 512 | 2x | 86.71 | 165.47 | 0.88 | 182.69 | 184.19 | 11,806.44 | 100.0% |
| 1,024 | 512 | 4x | 169.79 | 324.08 | 0.46 | 309.53 | 378.66 | 12,055.14 | 100.0% |
| 1,024 | 512 | 8x | 299.10 | 624.59 | 0.26 | 571.53 | 647.28 | 12,529.25 | 100.0% |
| 1,024 | 512 | 16x | 504.96 | 1,145.78 | 0.15 | 678.95 | 1,213.26 | 13,663.08 | 100.0% |
| 1,024 | 512 | 32x | 967.91 | 1,989.28 | 0.09 | 1,067.44 | 2,311.64 | 15,204.91 | 96.9% |
| 1,024 | 512 | 64x | 1,509.79 | 3,267.91 | 0.06 | 1,664.97 | 4,365.66 | 19,059.42 | 100.0% |
| 1,024 | 512 | 128x | 2,194.36 | 4,682.17 | 0.05 | 2,612.00 | 8,695.84 | 26,484.00 | 100.0% |
| 1,024 | 512 | 256x | 2,513.97 | 5,356.17 | 0.04 | 4,895.48 | 16,487.47 | 45,216.76 | 99.6% |
| 1,024 | 512 | 512x | 1,949.91 | 4,140.01 | 0.05 | 21,743.88 | 87,079.55 | 114,638.42 | 99.8% |
| 1,024 | 1,024 | 1x | 38.13 | 36.31 | 2.84 | 3,857.14 | 3,857.14 | 26,853.44 | 100.0% |
| 1,024 | 1,024 | 2x | 87.27 | 83.26 | 1.37 | 190.83 | 192.34 | 23,465.19 | 100.0% |
| 1,024 | 1,024 | 4x | 172.40 | 164.53 | 0.70 | 313.04 | 354.18 | 23,753.93 | 100.0% |
| 1,024 | 1,024 | 8x | 333.16 | 318.56 | 0.37 | 436.93 | 644.26 | 24,570.14 | 100.0% |
| 1,024 | 1,024 | 16x | 552.28 | 599.93 | 0.22 | 706.81 | 1,202.26 | 26,156.05 | 100.0% |
| 1,024 | 1,024 | 32x | 1,046.51 | 1,103.90 | 0.13 | 934.26 | 2,314.32 | 28,354.53 | 100.0% |
| 1,024 | 1,024 | 64x | 1,585.55 | 1,839.32 | 0.09 | 1,581.13 | 4,475.56 | 33,445.46 | 98.4% |
| 1,024 | 1,024 | 128x | 2,498.18 | 2,740.97 | 0.06 | 2,605.55 | 8,696.37 | 45,431.76 | 100.0% |
| 1,024 | 1,024 | 256x | 2,933.97 | 3,246.14 | 0.05 | 5,693.61 | 18,573.72 | 75,637.54 | 100.0% |
| 1,024 | 1,024 | 512x | 2,075.02 | 2,305.38 | 0.07 | 14,184.58 | 35,204.64 | 175,177.05 | 82.4% |
| 1,024 | 2,048 | 1x | 44.14 | 21.01 | 3.59 | 183.27 | 183.27 | 46,393.76 | 100.0% |
| 1,024 | 2,048 | 2x | 50.41 | 42.16 | 2.52 | 184.48 | 185.65 | 44,357.10 | 100.0% |
| 1,024 | 2,048 | 4x | 87.90 | 83.08 | 1.38 | 288.10 | 358.73 | 47,036.14 | 100.0% |
| 1,024 | 2,048 | 8x | 309.52 | 160.41 | 0.52 | 574.18 | 653.40 | 48,810.78 | 100.0% |
| 1,024 | 2,048 | 16x | 622.96 | 301.31 | 0.27 | 672.58 | 1,211.85 | 52,095.17 | 100.0% |
| 1,024 | 2,048 | 32x | 986.70 | 565.49 | 0.18 | 1,166.19 | 2,302.82 | 55,490.08 | 100.0% |
| 1,024 | 2,048 | 64x | 1,500.00 | 984.14 | 0.12 | 1,594.20 | 4,389.69 | 63,721.77 | 100.0% |
| 1,024 | 2,048 | 128x | 2,620.50 | 1,434.56 | 0.08 | 2,607.76 | 8,565.20 | 86,550.48 | 99.2% |
| 1,024 | 2,048 | 256x | 2,422.84 | 1,469.72 | 0.08 | 4,817.84 | 16,411.85 | 163,231.99 | 99.6% |
| 2,048 | 128 | 1x | 39.78 | 611.56 | 0.28 | 335.50 | 335.50 | 3,217.66 | 100.0% |
| 2,048 | 128 | 2x | 40.92 | 1,214.82 | 0.17 | 330.33 | 342.87 | 3,083.07 | 100.0% |
| 2,048 | 128 | 4x | 143.06 | 2,190.28 | 0.10 | 488.39 | 624.71 | 3,573.84 | 100.0% |
| 2,048 | 128 | 8x | 241.00 | 3,683.69 | 0.06 | 847.66 | 1,101.91 | 4,222.63 | 100.0% |
| 2,048 | 128 | 16x | 351.32 | 5,589.89 | 0.05 | 1,029.39 | 2,328.92 | 5,533.32 | 100.0% |
| 2,048 | 128 | 32x | 493.27 | 7,723.98 | 0.04 | 1,640.98 | 4,526.95 | 7,973.35 | 100.0% |
| 2,048 | 128 | 64x | 577.68 | 9,411.26 | 0.03 | 2,765.57 | 8,837.49 | 12,993.79 | 100.0% |
| 2,048 | 128 | 128x | 656.16 | 10,466.05 | 0.03 | 4,929.07 | 17,061.33 | 23,076.91 | 99.2% |
| 2,048 | 128 | 256x | 643.08 | 10,236.66 | 0.03 | 10,713.26 | 39,646.06 | 46,844.64 | 100.0% |
| 2,048 | 512 | 1x | 43.09 | 165.61 | 1.07 | 322.97 | 322.97 | 11,882.74 | 100.0% |
| 2,048 | 512 | 2x | 52.52 | 327.21 | 0.60 | 205.74 | 328.56 | 11,522.45 | 100.0% |
| 2,048 | 512 | 4x | 124.62 | 635.61 | 0.31 | 475.56 | 610.97 | 12,323.84 | 100.0% |
| 2,048 | 512 | 8x | 295.47 | 1,177.37 | 0.17 | 1,049.69 | 1,197.50 | 13,284.38 | 100.0% |
| 2,048 | 512 | 16x | 503.49 | 2,059.32 | 0.10 | 1,116.10 | 2,348.56 | 15,126.64 | 100.0% |
| 2,048 | 512 | 32x | 830.14 | 3,408.80 | 0.07 | 1,705.44 | 4,520.91 | 18,225.60 | 100.0% |
| 2,048 | 512 | 64x | 1,126.57 | 4,965.72 | 0.05 | 3,611.94 | 9,189.32 | 24,995.91 | 100.0% |
| 2,048 | 512 | 128x | 1,434.63 | 6,088.29 | 0.04 | 5,967.15 | 19,243.28 | 40,500.18 | 100.0% |
| 2,048 | 512 | 256x | 1,183.32 | 5,026.69 | 0.05 | 18,927.60 | 79,123.01 | 97,646.22 | 100.0% |
| 2,048 | 1,024 | 1x | 43.61 | 83.82 | 1.79 | 335.00 | 335.00 | 23,477.73 | 100.0% |
| 2,048 | 1,024 | 2x | 86.25 | 165.04 | 0.93 | 214.68 | 337.95 | 23,740.06 | 100.0% |
| 2,048 | 1,024 | 4x | 114.35 | 325.49 | 0.53 | 493.22 | 628.94 | 24,073.12 | 100.0% |
| 2,048 | 1,024 | 8x | 259.71 | 617.58 | 0.27 | 827.82 | 1,084.91 | 25,335.86 | 100.0% |
| 2,048 | 1,024 | 16x | 548.29 | 1,116.54 | 0.15 | 1,075.83 | 2,340.86 | 27,961.44 | 100.0% |
| 2,048 | 1,024 | 32x | 880.90 | 1,951.96 | 0.10 | 1,692.77 | 4,496.86 | 31,901.84 | 100.0% |
| 2,048 | 1,024 | 64x | 1,290.47 | 2,727.97 | 0.07 | 6,285.79 | 11,850.79 | 45,608.21 | 100.0% |
| 2,048 | 1,024 | 128x | 1,901.65 | 4,150.28 | 0.05 | 4,995.77 | 16,567.49 | 59,580.03 | 100.0% |
| 2,048 | 1,024 | 256x | 1,296.59 | 2,832.23 | 0.07 | 8,531.87 | 28,148.90 | 141,502.48 | 84.8% |
Runtime parameters used across all benchmark runs
