March 9, 2026 at 01:06 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 | 5,547.95★ | 1,388.05 | 0.01 | 84.01 | 125.76 | 9,123.15 | 100.0% |
| 128 | 128 | 1x | 86.84 | 86.84 | 0.29 | 37.70 | 37.70 | 1,474.61 | 100.0% |
| 128 | 128 | 2x | 171.01 | 169.00 | 0.15 | 22.14 | 22.63 | 1,496.49 | 100.0% |
| 128 | 128 | 4x | 333.99 | 332.03 | 0.08 | 40.76 | 45.98 | 1,531.33 | 100.0% |
| 128 | 128 | 8x | 121.90 | 120.36 | 0.13 | 6,912.73 | 6,914.80 | 8,397.22 | 100.0% |
| 128 | 512 | 1x | 88.17 | 22.04 | 0.65 | 19.63 | 19.63 | 5,807.12 | 100.0% |
| 128 | 512 | 2x | 171.70 | 42.42 | 0.34 | 22.80 | 23.17 | 5,963.30 | 100.0% |
| 128 | 512 | 4x | 338.68 | 84.17 | 0.17 | 31.41 | 35.67 | 6,044.00 | 100.0% |
| 128 | 512 | 8x | 673.67 | 167.30 | 0.09 | 33.81 | 36.61 | 6,039.33 | 100.0% |
| 128 | 512 | 16x | 1,267.78 | 320.28 | 0.05 | 77.98 | 100.22 | 6,372.76 | 100.0% |
| 128 | 512 | 32x | 2,302.81 | 582.48 | 0.03 | 103.76 | 176.08 | 7,017.05 | 100.0% |
| 128 | 512 | 64x | 4,043.79 | 1,030.61 | 0.02 | 149.77 | 316.77 | 7,913.26 | 100.0% |
| 128 | 512 | 128x | 5,353.03 | 1,354.83 | 0.02 | 184.38 | 372.09 | 9,364.86 | 100.0% |
| 128 | 512 | 256x | 5,507.07 | 1,391.06 | 0.02 | 89.86 | 118.18 | 9,103.71 | 100.0% |
| 128 | 512 | 1024x | 5,501.52 | 1,388.95 | 0.01 | 98.46 | 138.79 | 9,118.20 | 100.0% |
| 128 | 1,024 | 1x | 88.11 | 11.01 | 0.82 | 20.42 | 20.42 | 11,621.57 | 100.0% |
| 128 | 1,024 | 2x | 171.12 | 21.14 | 0.40 | 29.51 | 32.01 | 11,965.66 | 100.0% |
| 128 | 1,024 | 4x | 299.27 | 41.94 | 0.23 | 32.27 | 34.98 | 12,131.72 | 100.0% |
| 128 | 1,024 | 8x | 628.85 | 81.88 | 0.11 | 29.43 | 31.93 | 12,342.53 | 100.0% |
| 128 | 1,024 | 16x | 1,222.12 | 157.53 | 0.06 | 38.81 | 43.12 | 12,962.14 | 100.0% |
| 128 | 1,024 | 32x | 2,171.25 | 279.82 | 0.04 | 51.59 | 61.19 | 14,637.15 | 100.0% |
| 128 | 1,024 | 64x | 3,802.73 | 485.86 | 0.02 | 68.76 | 90.27 | 16,842.71 | 100.0% |
| 128 | 1,024 | 128x | 4,836.75 | 628.01 | 0.02 | 86.05 | 124.59 | 20,263.86 | 100.0% |
| 128 | 1,024 | 256x | 4,858.06 | 626.68 | 0.02 | 92.19 | 124.82 | 20,319.78 | 100.0% |
| 128 | 1,024 | 512x | 4,859.92 | 625.83 | 0.02 | 95.09 | 136.18 | 20,349.06 | 100.0% |
| 128 | 1,024 | 1024x | 4,857.65 | 626.35 | 0.02 | 94.59 | 138.94 | 20,328.76 | 100.0% |
| 128 | 2,048 | 1x | 87.53 | 5.47 | 0.91 | 20.02 | 20.02 | 23,397.14 | 100.0% |
| 128 | 2,048 | 2x | 169.63 | 10.48 | 0.45 | 22.27 | 22.74 | 24,145.74 | 100.0% |
| 128 | 2,048 | 4x | 273.06 | 20.71 | 0.28 | 30.05 | 32.07 | 24,569.88 | 100.0% |
| 128 | 2,048 | 8x | 570.18 | 39.55 | 0.14 | 31.14 | 33.67 | 25,556.56 | 100.0% |
| 128 | 2,048 | 16x | 1,123.76 | 74.82 | 0.07 | 41.14 | 46.85 | 27,306.58 | 100.0% |
| 128 | 2,048 | 32x | 1,861.67 | 128.76 | 0.05 | 54.65 | 65.28 | 31,844.37 | 100.0% |
| 128 | 2,048 | 64x | 3,099.51 | 216.35 | 0.03 | 66.87 | 91.35 | 37,901.00 | 100.0% |
| 128 | 2,048 | 128x | 3,854.18 | 268.41 | 0.03 | 89.22 | 122.90 | 47,560.64 | 100.0% |
| 128 | 2,048 | 256x | 3,876.65 | 266.25 | 0.03 | 132.84 | 209.02 | 47,880.19 | 100.0% |
| 128 | 2,048 | 512x | 3,876.41 | 267.85 | 0.03 | 105.25 | 135.43 | 47,562.84 | 100.0% |
| 512 | 128 | 1x | 86.20 | 333.33 | 0.13 | 45.16 | 45.16 | 1,484.97 | 100.0% |
| 512 | 128 | 2x | 167.76 | 649.41 | 0.07 | 45.85 | 46.26 | 1,525.97 | 100.0% |
| 512 | 128 | 4x | 321.20 | 1,248.43 | 0.03 | 88.77 | 89.77 | 1,593.06 | 100.0% |
| 512 | 128 | 8x | 609.89 | 2,360.33 | 0.02 | 115.87 | 163.49 | 1,675.66 | 100.0% |
| 512 | 128 | 16x | 1,084.75 | 4,192.80 | 0.01 | 178.45 | 287.92 | 1,878.68 | 100.0% |
| 512 | 128 | 32x | 1,740.02 | 6,734.49 | 0.01 | 294.88 | 551.35 | 2,328.99 | 100.0% |
| 512 | 128 | 64x | 2,618.93 | 10,144.82 | 0.01 | 541.83 | 1,067.42 | 3,086.36 | 100.0% |
| 512 | 128 | 128x | 3,428.88 | 13,300.03 | 0.01 | 493.18 | 1,228.15 | 3,659.14 | 100.0% |
| 512 | 128 | 256x | 4,760.13 | 18,463.74 | 0.00 | 140.78 | 187.51 | 2,608.37 | 100.0% |
| 512 | 128 | 512x | 4,760.13 | 18,463.74 | 0.00 | 147.20 | 197.67 | 2,610.96 | 100.0% |
| 512 | 128 | 1024x | 4,746.01 | 18,408.97 | 0.00 | 145.31 | 200.10 | 2,612.14 | 100.0% |
| 512 | 512 | 1x | 87.90 | 84.98 | 0.44 | 22.45 | 22.45 | 5,825.11 | 100.0% |
| 512 | 512 | 2x | 170.61 | 165.11 | 0.22 | 24.49 | 24.89 | 6,000.56 | 100.0% |
| 512 | 512 | 4x | 333.66 | 324.21 | 0.11 | 38.96 | 42.10 | 6,136.61 | 100.0% |
| 512 | 512 | 8x | 660.43 | 638.99 | 0.06 | 38.65 | 42.53 | 6,198.69 | 100.0% |
| 512 | 512 | 16x | 1,235.41 | 1,193.79 | 0.03 | 51.67 | 58.59 | 6,619.37 | 100.0% |
| 512 | 512 | 32x | 2,145.18 | 2,079.08 | 0.02 | 79.91 | 94.20 | 7,605.25 | 100.0% |
| 512 | 512 | 64x | 3,654.56 | 3,557.91 | 0.01 | 114.39 | 149.43 | 8,862.51 | 100.0% |
| 512 | 512 | 128x | 4,683.93 | 4,542.04 | 0.01 | 136.56 | 167.48 | 10,838.07 | 100.0% |
| 512 | 512 | 256x | 4,670.51 | 4,542.87 | 0.01 | 153.89 | 209.18 | 10,843.29 | 100.0% |
| 512 | 1,024 | 1x | 87.77 | 42.43 | 0.63 | 23.04 | 23.04 | 11,667.05 | 100.0% |
| 512 | 1,024 | 2x | 169.89 | 82.21 | 0.31 | 31.81 | 35.30 | 12,054.61 | 100.0% |
| 512 | 1,024 | 4x | 298.00 | 161.89 | 0.17 | 39.00 | 41.28 | 12,290.08 | 100.0% |
| 512 | 1,024 | 8x | 643.12 | 314.85 | 0.09 | 41.95 | 46.87 | 12,581.71 | 100.0% |
| 512 | 1,024 | 16x | 1,185.00 | 583.22 | 0.05 | 49.18 | 53.84 | 13,557.86 | 100.0% |
| 512 | 1,024 | 32x | 2,014.80 | 985.82 | 0.03 | 81.11 | 102.83 | 16,059.95 | 100.0% |
| 512 | 1,024 | 64x | 3,302.37 | 1,658.03 | 0.02 | 122.66 | 150.53 | 19,084.18 | 100.0% |
| 512 | 1,024 | 128x | 4,130.41 | 2,101.54 | 0.02 | 148.45 | 209.97 | 23,527.05 | 100.0% |
| 512 | 1,024 | 256x | 4,162.51 | 2,094.10 | 0.02 | 164.54 | 221.34 | 23,617.98 | 100.0% |
| 512 | 1,024 | 512x | 4,155.13 | 2,096.31 | 0.02 | 149.72 | 207.66 | 23,595.44 | 100.0% |
| 512 | 2,048 | 1x | 87.16 | 21.07 | 0.78 | 22.45 | 22.45 | 23,496.22 | 100.0% |
| 512 | 2,048 | 2x | 168.49 | 40.77 | 0.39 | 31.02 | 34.06 | 24,307.59 | 100.0% |
| 512 | 2,048 | 4x | 268.97 | 80.08 | 0.23 | 33.07 | 37.55 | 24,846.55 | 100.0% |
| 512 | 2,048 | 8x | 560.33 | 151.21 | 0.12 | 37.67 | 41.13 | 26,202.42 | 100.0% |
| 512 | 2,048 | 16x | 1,078.90 | 278.57 | 0.07 | 49.35 | 57.79 | 28,400.97 | 100.0% |
| 512 | 2,048 | 32x | 1,757.80 | 463.44 | 0.04 | 81.75 | 101.90 | 34,184.05 | 100.0% |
| 512 | 2,048 | 64x | 2,735.04 | 760.29 | 0.03 | 125.49 | 163.11 | 41,677.79 | 100.0% |
| 512 | 2,048 | 128x | 3,389.80 | 936.42 | 0.03 | 197.07 | 315.65 | 52,928.36 | 100.0% |
| 512 | 2,048 | 256x | 3,404.10 | 933.48 | 0.03 | 159.83 | 212.99 | 53,096.28 | 100.0% |
| 512 | 2,048 | 512x | 3,397.89 | 937.94 | 0.03 | 141.62 | 195.08 | 52,832.96 | 100.0% |
| 1,024 | 128 | 1x | 83.82 | 636.54 | 0.08 | 81.75 | 81.75 | 1,527.65 | 100.0% |
| 1,024 | 128 | 2x | 162.54 | 1,235.56 | 0.04 | 82.79 | 83.17 | 1,574.47 | 100.0% |
| 1,024 | 128 | 4x | 304.40 | 2,332.34 | 0.02 | 151.62 | 152.94 | 1,681.81 | 100.0% |
| 1,024 | 128 | 8x | 554.41 | 4,254.47 | 0.01 | 192.91 | 289.15 | 1,843.60 | 100.0% |
| 1,024 | 128 | 16x | 911.44 | 7,010.24 | 0.01 | 302.77 | 551.20 | 2,235.93 | 100.0% |
| 1,024 | 128 | 32x | 1,336.81 | 10,260.77 | 0.01 | 559.95 | 1,053.50 | 3,041.93 | 100.0% |
| 1,024 | 128 | 64x | 1,828.57 | 14,042.86 | 0.01 | 1,008.83 | 2,068.59 | 4,434.61 | 100.0% |
| 1,024 | 128 | 128x | 2,369.49 | 18,191.23 | 0.00 | 909.39 | 2,392.50 | 5,320.06 | 100.0% |
| 1,024 | 128 | 256x | 3,834.63 | 29,439.48 | 0.00 | 223.51 | 327.73 | 3,253.03 | 100.0% |
| 1,024 | 128 | 512x | 3,855.42 | 29,599.10 | 0.00 | 218.96 | 306.82 | 3,239.88 | 100.0% |
| 1,024 | 128 | 1024x | 3,820.90 | 29,334.03 | 0.00 | 220.70 | 327.86 | 3,271.18 | 100.0% |
| 1,024 | 512 | 1x | 87.48 | 166.07 | 0.30 | 23.93 | 23.93 | 5,853.22 | 100.0% |
| 1,024 | 512 | 2x | 169.20 | 321.55 | 0.15 | 28.12 | 28.44 | 6,051.57 | 100.0% |
| 1,024 | 512 | 4x | 328.31 | 628.89 | 0.08 | 42.33 | 46.24 | 6,235.62 | 100.0% |
| 1,024 | 512 | 8x | 625.82 | 1,225.13 | 0.04 | 48.48 | 53.55 | 6,409.45 | 100.0% |
| 1,024 | 512 | 16x | 1,153.17 | 2,242.28 | 0.02 | 67.62 | 77.92 | 7,010.57 | 100.0% |
| 1,024 | 512 | 32x | 1,897.10 | 3,688.72 | 0.02 | 94.52 | 123.11 | 8,495.69 | 100.0% |
| 1,024 | 512 | 64x | 3,079.26 | 6,036.46 | 0.01 | 163.16 | 215.21 | 10,373.35 | 100.0% |
| 1,024 | 512 | 128x | 3,852.27 | 7,437.30 | 0.01 | 216.74 | 310.70 | 13,125.36 | 100.0% |
| 1,024 | 512 | 256x | 3,844.56 | 7,447.44 | 0.01 | 219.71 | 308.36 | 13,107.70 | 100.0% |
| 1,024 | 1,024 | 1x | 87.19 | 108.10 | 0.41 | 24.58 | 24.58 | 8,980.59 | 100.0% |
| 1,024 | 1,024 | 2x | 136.53 | 162.30 | 0.27 | 34.76 | 38.60 | 11,752.48 | 100.0% |
| 1,024 | 1,024 | 4x | 295.15 | 314.29 | 0.13 | 34.36 | 35.66 | 12,479.66 | 100.0% |
| 1,024 | 1,024 | 8x | 586.51 | 600.81 | 0.07 | 56.49 | 70.81 | 13,074.14 | 100.0% |
| 1,024 | 1,024 | 16x | 1,052.48 | 1,109.61 | 0.04 | 66.17 | 81.05 | 14,184.65 | 100.0% |
| 1,024 | 1,024 | 32x | 1,762.37 | 1,769.52 | 0.03 | 106.04 | 132.40 | 17,741.33 | 100.0% |
| 1,024 | 1,024 | 64x | 2,830.12 | 2,849.28 | 0.02 | 164.03 | 225.42 | 22,024.95 | 100.0% |
| 1,024 | 1,024 | 128x | 3,471.25 | 3,511.74 | 0.02 | 222.90 | 295.82 | 27,896.97 | 100.0% |
| 1,024 | 1,024 | 256x | 3,487.41 | 3,501.23 | 0.02 | 215.45 | 309.50 | 27,974.78 | 100.0% |
| 1,024 | 1,024 | 512x | 3,473.12 | 3,505.48 | 0.02 | 238.31 | 320.04 | 27,950.20 | 100.0% |
| 1,024 | 2,048 | 1x | 87.25 | 185.57 | 0.28 | 25.97 | 25.97 | 5,226.10 | 100.0% |
| 1,024 | 2,048 | 2x | 114.60 | 100.55 | 0.39 | 28.31 | 28.71 | 18,702.80 | 100.0% |
| 1,024 | 2,048 | 4x | 246.01 | 157.05 | 0.20 | 35.26 | 38.37 | 24,965.16 | 100.0% |
| 1,024 | 2,048 | 8x | 496.94 | 297.29 | 0.11 | 48.67 | 58.66 | 26,105.49 | 100.0% |
| 1,024 | 2,048 | 16x | 918.44 | 544.35 | 0.06 | 67.15 | 83.83 | 28,914.26 | 100.0% |
| 1,024 | 2,048 | 32x | 1,529.16 | 866.92 | 0.04 | 108.04 | 135.93 | 36,241.00 | 100.0% |
| 1,024 | 2,048 | 64x | 2,403.05 | 1,336.11 | 0.03 | 157.27 | 205.22 | 47,028.94 | 100.0% |
| 1,024 | 2,048 | 128x | 2,909.65 | 1,612.32 | 0.02 | 313.47 | 613.40 | 60,857.59 | 100.0% |
| 1,024 | 2,048 | 256x | 2,910.68 | 1,623.26 | 0.02 | 226.45 | 325.64 | 60,448.14 | 100.0% |
| 1,024 | 2,048 | 512x | 2,938.13 | 1,643.18 | 0.02 | 219.71 | 302.43 | 59,714.41 | 100.0% |
| 1,024 | 2,048 | 1024x | 2,909.97 | 1,629.34 | 0.02 | 229.72 | 322.48 | 60,217.70 | 100.0% |
| 2,048 | 128 | 1x | 79.31 | 1,209.42 | 0.04 | 148.05 | 148.05 | 1,614.24 | 100.0% |
| 2,048 | 128 | 2x | 152.11 | 2,317.88 | 0.02 | 107.29 | 158.34 | 1,681.81 | 100.0% |
| 2,048 | 128 | 4x | 274.09 | 4,179.34 | 0.01 | 233.72 | 292.55 | 1,865.93 | 100.0% |
| 2,048 | 128 | 8x | 461.68 | 7,046.44 | 0.01 | 496.99 | 565.72 | 2,215.87 | 100.0% |
| 2,048 | 128 | 16x | 693.30 | 10,585.99 | 0.01 | 948.97 | 1,084.16 | 2,949.78 | 100.0% |
| 2,048 | 128 | 32x | 903.40 | 13,798.19 | 0.01 | 1,286.29 | 2,104.54 | 4,515.01 | 100.0% |
| 2,048 | 128 | 64x | 1,120.81 | 17,122.45 | 0.01 | 1,828.83 | 4,027.67 | 7,249.50 | 100.0% |
| 2,048 | 128 | 128x | 1,440.79 | 22,015.98 | 0.00 | 1,681.78 | 4,856.82 | 8,796.40 | 100.0% |
| 2,048 | 128 | 256x | 2,758.03 | 42,143.93 | 0.00 | 387.88 | 600.76 | 4,555.05 | 100.0% |
| 2,048 | 128 | 512x | 2,782.61 | 42,519.57 | 0.00 | 368.29 | 563.23 | 4,511.69 | 100.0% |
| 2,048 | 128 | 1024x | 2,781.40 | 42,501.09 | 0.00 | 385.32 | 570.98 | 4,515.25 | 100.0% |
| 2,048 | 512 | 1x | 86.15 | 328.45 | 0.19 | 33.57 | 33.57 | 5,943.47 | 100.0% |
| 2,048 | 512 | 2x | 166.15 | 632.97 | 0.09 | 33.90 | 34.29 | 6,162.27 | 100.0% |
| 2,048 | 512 | 4x | 318.95 | 1,215.85 | 0.05 | 48.39 | 49.49 | 6,418.79 | 100.0% |
| 2,048 | 512 | 8x | 602.18 | 2,297.71 | 0.03 | 74.14 | 83.00 | 6,798.35 | 100.0% |
| 2,048 | 512 | 16x | 1,050.26 | 4,009.10 | 0.02 | 111.38 | 129.99 | 7,792.68 | 100.0% |
| 2,048 | 512 | 32x | 1,574.94 | 6,034.05 | 0.01 | 160.60 | 205.90 | 10,344.69 | 100.0% |
| 2,048 | 512 | 64x | 2,418.48 | 9,236.70 | 0.01 | 262.27 | 367.51 | 13,495.27 | 100.0% |
| 2,048 | 512 | 128x | 2,864.12 | 11,014.19 | 0.01 | 387.74 | 580.76 | 17,669.75 | 100.0% |
| 2,048 | 512 | 256x | 2,859.07 | 11,016.67 | 0.01 | 362.64 | 537.01 | 17,638.09 | 100.0% |
| 2,048 | 1,024 | 1x | 86.10 | 164.13 | 0.33 | 33.33 | 33.33 | 11,892.45 | 100.0% |
| 2,048 | 1,024 | 2x | 165.84 | 315.90 | 0.16 | 33.61 | 34.00 | 12,347.96 | 100.0% |
| 2,048 | 1,024 | 4x | 317.45 | 605.05 | 0.09 | 46.32 | 47.57 | 12,901.18 | 100.0% |
| 2,048 | 1,024 | 8x | 597.08 | 1,139.14 | 0.05 | 65.37 | 81.64 | 13,714.43 | 100.0% |
| 2,048 | 1,024 | 16x | 1,033.50 | 1,972.56 | 0.03 | 110.71 | 126.32 | 15,843.39 | 100.0% |
| 2,048 | 1,024 | 32x | 1,494.97 | 2,916.32 | 0.02 | 175.14 | 219.40 | 21,427.49 | 100.0% |
| 2,048 | 1,024 | 64x | 2,273.10 | 4,438.19 | 0.02 | 276.12 | 375.71 | 28,142.55 | 100.0% |
| 2,048 | 1,024 | 128x | 2,672.86 | 5,257.23 | 0.01 | 453.21 | 826.74 | 37,109.87 | 100.0% |
| 2,048 | 1,024 | 256x | 2,663.27 | 5,322.47 | 0.01 | 392.23 | 570.57 | 36,660.50 | 100.0% |
| 2,048 | 2,048 | 1x | 85.69 | 81.67 | 0.51 | 30.85 | 30.85 | 23,901.04 | 100.0% |
| 2,048 | 2,048 | 2x | 160.28 | 169.72 | 0.25 | 34.20 | 34.68 | 22,906.78 | 100.0% |
| 2,048 | 2,048 | 4x | 276.07 | 303.56 | 0.14 | 48.03 | 54.97 | 25,318.85 | 100.0% |
| 2,048 | 2,048 | 8x | 553.66 | 558.16 | 0.08 | 72.75 | 81.82 | 27,993.72 | 100.0% |
| 2,048 | 2,048 | 16x | 947.44 | 964.56 | 0.05 | 93.34 | 113.13 | 32,405.61 | 100.0% |
| 2,048 | 2,048 | 32x | 1,339.21 | 1,444.73 | 0.03 | 175.49 | 216.30 | 43,276.64 | 100.0% |
| 2,048 | 2,048 | 64x | 1,967.34 | 2,105.81 | 0.03 | 662.56 | 1,134.26 | 59,367.86 | 100.0% |
| 2,048 | 2,048 | 128x | 2,216.63 | 2,424.66 | 0.02 | 1,768.51 | 4,713.15 | 80,575.73 | 100.0% |
| 2,048 | 2,048 | 256x | 2,306.96 | 2,542.38 | 0.02 | 884.50 | 1,638.60 | 76,841.94 | 100.0% |
| 2,048 | 2,048 | 512x | 2,304.74 | 2,553.33 | 0.02 | 834.35 | 1,481.76 | 76,516.07 | 100.0% |
Runtime parameters used across all benchmark runs
