November 6, 2025 at 07:07 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 | 1024x | 11,481.64★ | 3,305.38 | 0.02 | 1,004.27 | 2,648.07 | 11,446.45 | 100.0% |
| 128 | 128 | 1x | 79.60 | 75.87 | 0.40 | 846.69 | 846.69 | 1,607.79 | 100.0% |
| 128 | 128 | 2x | 163.06 | 159.24 | 0.22 | 774.05 | 775.74 | 1,565.28 | 100.0% |
| 128 | 128 | 4x | 582.48 | 577.93 | 0.06 | 86.34 | 112.44 | 871.50 | 100.0% |
| 128 | 128 | 8x | 1,015.87 | 1,018.85 | 0.06 | 132.10 | 173.28 | 990.56 | 100.0% |
| 128 | 128 | 16x | 1,935.73 | 1,949.91 | 0.03 | 151.22 | 203.53 | 1,029.41 | 100.0% |
| 128 | 128 | 32x | 3,429.65 | 3,435.51 | 0.02 | 163.17 | 205.18 | 1,141.07 | 100.0% |
| 128 | 128 | 64x | 4,186.57 | 4,487.62 | 0.02 | 156.54 | 237.84 | 1,246.44 | 100.0% |
| 128 | 128 | 128x | 5,083.06 | 6,046.14 | 0.02 | 240.47 | 451.22 | 1,464.58 | 100.0% |
| 128 | 128 | 256x | 6,132.06 | 7,106.95 | 0.01 | 253.04 | 464.98 | 1,598.83 | 100.0% |
| 128 | 128 | 512x | 5,560.85 | 7,362.31 | 0.02 | 441.00 | 1,591.19 | 3,286.10 | 100.0% |
| 128 | 512 | 1x | 170.78 | 40.69 | 0.55 | 46.75 | 46.75 | 2,997.98 | 100.0% |
| 128 | 512 | 2x | 324.97 | 79.42 | 0.36 | 53.48 | 59.09 | 3,142.63 | 100.0% |
| 128 | 512 | 4x | 626.87 | 155.49 | 0.21 | 73.73 | 85.14 | 3,256.42 | 100.0% |
| 128 | 512 | 8x | 838.20 | 225.47 | 0.13 | 90.45 | 102.63 | 4,548.67 | 100.0% |
| 128 | 512 | 16x | 1,481.67 | 400.19 | 0.07 | 128.75 | 160.56 | 5,106.86 | 100.0% |
| 128 | 512 | 32x | 3,926.31 | 1,000.98 | 0.03 | 150.17 | 178.76 | 4,032.41 | 100.0% |
| 128 | 512 | 64x | 6,246.53 | 1,642.30 | 0.02 | 142.02 | 196.06 | 4,519.35 | 100.0% |
| 128 | 512 | 128x | 7,847.36 | 2,062.85 | 0.02 | 202.18 | 411.53 | 5,401.76 | 100.0% |
| 128 | 512 | 256x | 9,466.91 | 2,555.82 | 0.02 | 288.83 | 620.58 | 6,642.71 | 100.0% |
| 128 | 512 | 512x | 10,702.51 | 2,900.87 | 0.02 | 955.30 | 2,788.34 | 12,366.08 | 100.0% |
| 128 | 1,024 | 1x | 157.04 | 19.63 | 0.62 | 56.24 | 56.24 | 6,205.59 | 100.0% |
| 128 | 1,024 | 2x | 254.28 | 39.61 | 0.48 | 51.83 | 55.96 | 6,172.50 | 100.0% |
| 128 | 1,024 | 4x | 471.42 | 74.65 | 0.21 | 93.87 | 98.56 | 6,659.77 | 100.0% |
| 128 | 1,024 | 8x | 711.82 | 151.54 | 0.18 | 114.24 | 120.20 | 6,652.21 | 100.0% |
| 128 | 1,024 | 16x | 1,596.77 | 284.79 | 0.09 | 161.91 | 193.95 | 7,184.86 | 100.0% |
| 128 | 1,024 | 32x | 3,066.49 | 520.49 | 0.05 | 165.15 | 191.62 | 7,813.03 | 100.0% |
| 128 | 1,024 | 64x | 5,070.11 | 862.22 | 0.04 | 152.46 | 208.36 | 9,324.32 | 100.0% |
| 128 | 1,024 | 128x | 6,990.34 | 1,172.11 | 0.03 | 189.01 | 414.60 | 11,146.71 | 100.0% |
| 128 | 1,024 | 256x | 8,977.95 | 1,530.61 | 0.02 | 304.19 | 722.33 | 14,216.69 | 100.0% |
| 128 | 1,024 | 512x | 10,492.76 | 1,800.59 | 0.02 | 627.96 | 1,444.98 | 18,565.60 | 100.0% |
| 128 | 1,024 | 1024x | 11,322.90 | 1,922.74 | 0.02 | 1,058.05 | 2,398.37 | 27,024.78 | 100.0% |
| 128 | 2,048 | 1x | 172.00 | 23.82 | 0.59 | 57.38 | 57.38 | 5,115.29 | 100.0% |
| 128 | 2,048 | 2x | 323.62 | 59.23 | 0.28 | 63.70 | 65.08 | 4,198.09 | 100.0% |
| 128 | 2,048 | 4x | 508.78 | 79.62 | 0.28 | 62.14 | 65.95 | 6,294.49 | 100.0% |
| 128 | 2,048 | 8x | 719.65 | 145.49 | 0.17 | 110.68 | 138.70 | 6,272.11 | 100.0% |
| 128 | 2,048 | 16x | 1,309.22 | 250.85 | 0.10 | 142.71 | 188.30 | 7,097.18 | 100.0% |
| 128 | 2,048 | 32x | 2,289.52 | 369.28 | 0.07 | 149.28 | 173.28 | 9,360.21 | 100.0% |
| 128 | 2,048 | 64x | 4,110.64 | 614.98 | 0.04 | 153.94 | 195.62 | 12,638.20 | 100.0% |
| 128 | 2,048 | 128x | 4,651.90 | 740.46 | 0.04 | 161.18 | 272.48 | 14,255.68 | 100.0% |
| 128 | 2,048 | 256x | 7,274.34 | 1,153.57 | 0.03 | 561.30 | 2,724.37 | 20,094.99 | 100.0% |
| 128 | 2,048 | 512x | 9,036.25 | 1,429.96 | 0.02 | 601.07 | 1,791.17 | 24,934.97 | 100.0% |
| 128 | 2,048 | 1024x | 10,987.18 | 1,722.33 | 0.02 | 817.88 | 2,594.79 | 32,590.06 | 100.0% |
| 512 | 128 | 1x | 155.72 | 603.41 | 0.08 | 52.41 | 52.41 | 821.66 | 100.0% |
| 512 | 128 | 2x | 298.72 | 1,161.03 | 0.05 | 69.26 | 89.05 | 851.64 | 100.0% |
| 512 | 128 | 4x | 568.89 | 2,197.78 | 0.05 | 77.52 | 87.29 | 892.11 | 100.0% |
| 512 | 128 | 8x | 1,006.88 | 3,883.97 | 0.03 | 110.38 | 138.80 | 1,000.47 | 100.0% |
| 512 | 128 | 16x | 1,727.67 | 6,949.21 | 0.02 | 154.12 | 181.20 | 1,121.85 | 100.0% |
| 512 | 128 | 32x | 2,960.29 | 11,736.03 | 0.01 | 224.54 | 287.16 | 1,318.03 | 100.0% |
| 512 | 128 | 64x | 4,544.28 | 17,864.35 | 0.01 | 213.80 | 366.25 | 1,678.86 | 100.0% |
| 512 | 128 | 128x | 5,363.54 | 21,607.60 | 0.01 | 184.10 | 292.07 | 2,112.03 | 100.0% |
| 512 | 128 | 256x | 6,250.30 | 25,110.06 | 0.01 | 223.64 | 364.76 | 2,590.08 | 100.0% |
| 512 | 128 | 512x | 6,620.15 | 27,143.50 | 0.01 | 355.70 | 907.23 | 4,345.16 | 100.0% |
| 512 | 128 | 1024x | 7,549.57 | 30,444.45 | 0.01 | 380.31 | 728.15 | 4,002.21 | 100.0% |
| 512 | 512 | 1x | 168.53 | 163.27 | 0.32 | 46.97 | 46.97 | 3,036.84 | 100.0% |
| 512 | 512 | 2x | 330.00 | 320.66 | 0.24 | 45.49 | 45.78 | 3,097.91 | 100.0% |
| 512 | 512 | 4x | 547.47 | 601.95 | 0.14 | 85.55 | 105.74 | 3,271.22 | 100.0% |
| 512 | 512 | 8x | 1,032.63 | 1,111.11 | 0.08 | 106.11 | 129.13 | 3,544.04 | 100.0% |
| 512 | 512 | 16x | 1,569.47 | 1,937.97 | 0.05 | 171.14 | 219.95 | 4,070.94 | 100.0% |
| 512 | 512 | 32x | 3,103.38 | 3,306.61 | 0.03 | 279.22 | 382.96 | 4,783.49 | 100.0% |
| 512 | 512 | 64x | 5,451.40 | 5,736.14 | 0.02 | 216.57 | 257.44 | 5,435.62 | 100.0% |
| 512 | 512 | 128x | 7,884.26 | 8,034.33 | 0.02 | 212.82 | 315.42 | 7,677.98 | 100.0% |
| 512 | 512 | 256x | 7,171.22 | 7,376.75 | 0.02 | 603.01 | 2,233.35 | 11,673.92 | 100.0% |
| 512 | 1,024 | 1x | 164.05 | 235.85 | 0.28 | 49.87 | 49.87 | 2,096.81 | 100.0% |
| 512 | 1,024 | 2x | 200.94 | 161.37 | 0.36 | 50.22 | 53.76 | 5,925.69 | 100.0% |
| 512 | 1,024 | 4x | 438.73 | 313.17 | 0.16 | 97.99 | 102.34 | 6,043.87 | 100.0% |
| 512 | 1,024 | 8x | 932.04 | 766.99 | 0.10 | 102.81 | 125.03 | 4,819.54 | 100.0% |
| 512 | 1,024 | 16x | 1,141.95 | 933.32 | 0.08 | 125.09 | 146.26 | 7,673.13 | 100.0% |
| 512 | 1,024 | 32x | 2,441.57 | 1,911.04 | 0.04 | 153.50 | 179.55 | 7,647.95 | 100.0% |
| 512 | 1,024 | 64x | 3,992.89 | 3,105.33 | 0.03 | 145.64 | 222.94 | 9,657.40 | 100.0% |
| 512 | 1,024 | 128x | 6,277.75 | 4,495.48 | 0.02 | 171.41 | 323.49 | 12,502.55 | 100.0% |
| 512 | 1,024 | 256x | 7,622.53 | 5,710.67 | 0.02 | 261.27 | 624.85 | 17,369.64 | 100.0% |
| 512 | 1,024 | 512x | 8,428.06 | 6,048.08 | 0.02 | 498.52 | 1,507.71 | 25,677.13 | 100.0% |
| 512 | 1,024 | 1024x | 9,095.47 | 6,563.47 | 0.02 | 2,342.22 | 11,292.16 | 45,640.48 | 100.0% |
| 512 | 2,048 | 1x | 164.69 | 210.53 | 0.30 | 62.09 | 62.09 | 2,348.53 | 100.0% |
| 512 | 2,048 | 2x | 264.59 | 238.90 | 0.21 | 77.54 | 82.40 | 4,100.13 | 100.0% |
| 512 | 2,048 | 4x | 508.93 | 371.87 | 0.17 | 118.17 | 150.28 | 5,253.40 | 100.0% |
| 512 | 2,048 | 8x | 811.87 | 565.09 | 0.11 | 149.87 | 166.25 | 6,699.02 | 100.0% |
| 512 | 2,048 | 16x | 1,334.48 | 1,135.82 | 0.07 | 181.74 | 232.67 | 6,752.33 | 100.0% |
| 512 | 2,048 | 32x | 1,509.43 | 1,041.23 | 0.06 | 224.30 | 293.21 | 10,139.93 | 100.0% |
| 512 | 2,048 | 64x | 2,993.93 | 2,172.61 | 0.04 | 243.75 | 334.38 | 11,410.53 | 100.0% |
| 512 | 2,048 | 128x | 4,225.96 | 2,895.82 | 0.03 | 186.96 | 281.88 | 15,446.53 | 100.0% |
| 512 | 2,048 | 256x | 6,084.62 | 4,188.59 | 0.02 | 259.27 | 574.54 | 19,721.13 | 100.0% |
| 512 | 2,048 | 512x | 7,602.99 | 5,094.06 | 0.02 | 440.74 | 1,049.70 | 30,827.68 | 100.0% |
| 512 | 2,048 | 1024x | 8,551.50 | 5,744.01 | 0.02 | 3,175.27 | 14,443.09 | 53,117.24 | 100.0% |
| 1,024 | 128 | 1x | 145.12 | 1,105.44 | 0.05 | 91.99 | 91.99 | 881.63 | 100.0% |
| 1,024 | 128 | 2x | 286.35 | 2,185.68 | 0.03 | 65.58 | 80.90 | 888.68 | 100.0% |
| 1,024 | 128 | 4x | 527.84 | 4,029.90 | 0.03 | 109.51 | 136.10 | 968.02 | 100.0% |
| 1,024 | 128 | 8x | 948.15 | 7,252.78 | 0.02 | 123.77 | 142.19 | 1,068.95 | 100.0% |
| 1,024 | 128 | 16x | 1,586.37 | 12,169.64 | 0.01 | 218.64 | 256.46 | 1,275.52 | 100.0% |
| 1,024 | 128 | 32x | 2,419.87 | 18,584.86 | 0.01 | 274.05 | 400.64 | 1,636.36 | 100.0% |
| 1,024 | 128 | 64x | 3,429.05 | 26,408.06 | 0.01 | 349.74 | 549.53 | 2,285.79 | 100.0% |
| 1,024 | 128 | 128x | 4,592.20 | 35,806.49 | 0.01 | 390.02 | 691.84 | 3,308.66 | 100.0% |
| 1,024 | 128 | 256x | 4,934.48 | 39,160.75 | 0.01 | 341.97 | 800.28 | 4,482.06 | 100.0% |
| 1,024 | 128 | 512x | 5,030.18 | 39,368.60 | 0.01 | 672.07 | 1,936.57 | 9,145.26 | 100.0% |
| 1,024 | 128 | 1024x | 4,802.19 | 37,235.42 | 0.01 | 7,364.70 | 15,463.55 | 20,710.74 | 100.0% |
| 1,024 | 512 | 1x | 157.30 | 346.98 | 0.20 | 79.70 | 79.70 | 2,803.20 | 100.0% |
| 1,024 | 512 | 2x | 308.62 | 591.23 | 0.17 | 73.77 | 80.03 | 3,290.69 | 100.0% |
| 1,024 | 512 | 4x | 590.03 | 1,126.19 | 0.10 | 86.89 | 105.90 | 3,457.94 | 100.0% |
| 1,024 | 512 | 8x | 1,021.14 | 1,971.56 | 0.05 | 150.01 | 186.47 | 3,961.01 | 100.0% |
| 1,024 | 512 | 16x | 1,831.02 | 3,511.62 | 0.03 | 201.14 | 274.45 | 4,456.23 | 100.0% |
| 1,024 | 512 | 32x | 2,962.47 | 5,867.63 | 0.02 | 292.71 | 384.63 | 5,308.54 | 100.0% |
| 1,024 | 512 | 64x | 4,339.53 | 8,796.53 | 0.02 | 344.01 | 521.88 | 7,056.72 | 100.0% |
| 1,024 | 512 | 128x | 5,659.83 | 11,580.67 | 0.02 | 375.92 | 771.46 | 10,641.76 | 100.0% |
| 1,024 | 512 | 256x | 6,978.59 | 14,405.35 | 0.01 | 305.21 | 738.13 | 15,509.89 | 100.0% |
| 1,024 | 512 | 512x | 6,878.85 | 13,983.36 | 0.01 | 1,020.90 | 4,475.27 | 22,512.72 | 100.0% |
| 1,024 | 1,024 | 1x | 150.35 | 234.54 | 0.32 | 73.04 | 73.04 | 4,148.88 | 100.0% |
| 1,024 | 1,024 | 2x | 285.77 | 379.86 | 0.17 | 78.04 | 92.46 | 5,128.08 | 100.0% |
| 1,024 | 1,024 | 4x | 473.62 | 597.80 | 0.16 | 114.54 | 151.06 | 6,507.83 | 100.0% |
| 1,024 | 1,024 | 8x | 899.02 | 1,448.68 | 0.08 | 122.45 | 145.18 | 5,381.90 | 100.0% |
| 1,024 | 1,024 | 16x | 1,308.35 | 1,902.06 | 0.06 | 202.23 | 250.80 | 7,549.92 | 100.0% |
| 1,024 | 1,024 | 32x | 2,351.07 | 3,240.23 | 0.04 | 262.16 | 377.19 | 9,614.02 | 100.0% |
| 1,024 | 1,024 | 64x | 3,673.86 | 5,470.86 | 0.03 | 318.01 | 488.27 | 11,108.98 | 100.0% |
| 1,024 | 1,024 | 128x | 4,989.30 | 7,357.27 | 0.02 | 412.30 | 766.81 | 16,808.61 | 100.0% |
| 1,024 | 1,024 | 256x | 6,212.13 | 9,145.48 | 0.02 | 377.66 | 826.25 | 25,940.52 | 100.0% |
| 1,024 | 1,024 | 512x | 6,484.06 | 9,546.37 | 0.02 | 4,654.48 | 29,567.41 | 47,767.38 | 100.0% |
| 1,024 | 1,024 | 1024x | 6,294.79 | 9,255.38 | 0.02 | 28,559.17 | 68,209.11 | 93,701.66 | 100.0% |
| 1,024 | 2,048 | 1x | 167.69 | 259.93 | 0.27 | 83.33 | 83.33 | 3,743.82 | 100.0% |
| 1,024 | 2,048 | 2x | 294.70 | 386.47 | 0.23 | 86.08 | 94.13 | 5,008.42 | 100.0% |
| 1,024 | 2,048 | 4x | 545.17 | 874.11 | 0.11 | 94.09 | 107.51 | 4,397.39 | 100.0% |
| 1,024 | 2,048 | 8x | 907.13 | 1,332.37 | 0.08 | 141.04 | 161.99 | 5,653.59 | 100.0% |
| 1,024 | 2,048 | 16x | 1,063.12 | 1,437.29 | 0.07 | 212.43 | 287.74 | 9,531.88 | 100.0% |
| 1,024 | 2,048 | 32x | 2,178.97 | 2,906.68 | 0.04 | 275.74 | 434.44 | 9,793.31 | 100.0% |
| 1,024 | 2,048 | 64x | 2,752.69 | 3,598.63 | 0.04 | 366.61 | 565.01 | 12,822.68 | 100.0% |
| 1,024 | 2,048 | 128x | 3,642.88 | 5,340.35 | 0.03 | 363.41 | 790.26 | 17,237.66 | 100.0% |
| 1,024 | 2,048 | 256x | 5,107.52 | 7,182.94 | 0.02 | 526.40 | 1,249.08 | 27,332.66 | 100.0% |
| 1,024 | 2,048 | 512x | 5,634.46 | 7,851.74 | 0.02 | 4,843.06 | 31,115.90 | 53,216.70 | 100.0% |
| 1,024 | 2,048 | 1024x | 6,187.18 | 8,465.40 | 0.02 | 31,226.53 | 75,435.23 | 100,768.48 | 100.0% |
| 2,048 | 128 | 1x | 144.47 | 2,221.22 | 0.03 | 108.97 | 108.97 | 886.12 | 100.0% |
| 2,048 | 128 | 2x | 262.30 | 4,015.37 | 0.02 | 84.44 | 106.71 | 964.85 | 100.0% |
| 2,048 | 128 | 4x | 458.78 | 7,024.19 | 0.02 | 138.24 | 168.41 | 1,109.33 | 100.0% |
| 2,048 | 128 | 8x | 842.11 | 12,871.71 | 0.01 | 180.00 | 244.16 | 1,195.83 | 100.0% |
| 2,048 | 128 | 16x | 1,260.95 | 20,458.47 | 0.01 | 301.50 | 409.77 | 1,495.79 | 100.0% |
| 2,048 | 128 | 32x | 1,821.53 | 28,781.97 | 0.01 | 421.28 | 624.37 | 2,137.44 | 100.0% |
| 2,048 | 128 | 64x | 2,439.14 | 37,798.55 | 0.01 | 614.12 | 1,095.26 | 3,226.38 | 100.0% |
| 2,048 | 128 | 128x | 2,917.15 | 45,184.12 | 0.01 | 897.10 | 1,908.00 | 5,278.90 | 100.0% |
| 2,048 | 128 | 256x | 2,841.37 | 44,139.33 | 0.01 | 1,904.89 | 7,665.57 | 9,992.67 | 100.0% |
| 2,048 | 512 | 1x | 161.03 | 627.55 | 0.14 | 102.93 | 102.93 | 3,129.51 | 100.0% |
| 2,048 | 512 | 2x | 224.18 | 1,208.45 | 0.08 | 115.43 | 117.38 | 3,152.09 | 100.0% |
| 2,048 | 512 | 4x | 511.12 | 2,126.70 | 0.06 | 132.46 | 167.00 | 3,633.86 | 100.0% |
| 2,048 | 512 | 8x | 785.78 | 3,746.29 | 0.04 | 196.55 | 289.56 | 4,162.25 | 100.0% |
| 2,048 | 512 | 16x | 1,366.73 | 6,856.86 | 0.02 | 317.07 | 508.30 | 4,546.62 | 100.0% |
| 2,048 | 512 | 32x | 2,289.72 | 10,133.12 | 0.02 | 421.14 | 651.08 | 6,104.96 | 100.0% |
| 2,048 | 512 | 64x | 3,253.40 | 13,628.55 | 0.02 | 475.48 | 906.22 | 9,064.97 | 100.0% |
| 2,048 | 512 | 128x | 4,004.68 | 16,749.41 | 0.02 | 903.97 | 1,899.87 | 14,727.57 | 100.0% |
| 2,048 | 512 | 256x | 3,762.59 | 15,831.12 | 0.01 | 3,521.68 | 22,285.29 | 30,216.11 | 100.0% |
| 2,048 | 1,024 | 1x | 154.99 | 770.25 | 0.13 | 107.55 | 107.55 | 2,548.35 | 100.0% |
| 2,048 | 1,024 | 2x | 291.46 | 851.77 | 0.14 | 76.43 | 96.11 | 4,559.88 | 100.0% |
| 2,048 | 1,024 | 4x | 451.45 | 1,349.69 | 0.09 | 122.16 | 194.46 | 5,770.78 | 100.0% |
| 2,048 | 1,024 | 8x | 698.23 | 2,765.37 | 0.05 | 178.65 | 244.08 | 5,378.37 | 100.0% |
| 2,048 | 1,024 | 16x | 976.40 | 3,432.95 | 0.04 | 261.07 | 369.03 | 8,101.61 | 100.0% |
| 2,048 | 1,024 | 32x | 1,764.45 | 5,786.74 | 0.03 | 435.28 | 757.73 | 10,115.20 | 100.0% |
| 2,048 | 1,024 | 64x | 2,386.72 | 7,919.95 | 0.02 | 650.42 | 1,389.14 | 15,595.77 | 100.0% |
| 2,048 | 1,024 | 128x | 3,604.32 | 10,737.82 | 0.02 | 851.36 | 1,876.05 | 22,798.14 | 100.0% |
| 2,048 | 1,024 | 256x | 3,988.14 | 12,243.00 | 0.02 | 3,713.05 | 22,964.52 | 37,260.50 | 100.0% |
| 2,048 | 1,024 | 512x | 3,925.93 | 12,408.70 | 0.02 | 22,465.78 | 54,663.07 | 73,676.78 | 100.0% |
| 2,048 | 2,048 | 1x | 146.54 | 396.14 | 0.20 | 109.57 | 109.57 | 4,959.86 | 100.0% |
| 2,048 | 2,048 | 2x | 229.77 | 961.01 | 0.12 | 98.33 | 102.14 | 3,972.07 | 100.0% |
| 2,048 | 2,048 | 4x | 324.45 | 1,016.96 | 0.11 | 89.44 | 103.04 | 5,548.85 | 75.0% |
Runtime parameters used across all benchmark runs
