November 13, 2025 at 08:07 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 | |||||||||
| 2,048 | 2,048 | 1024x | 18,672.47★ | 21,871.30 | 0.03 | 14,591.20 | 21,056.79 | 82,870.74 | 100.0% |
| 128 | 512 | 1x | 31.22 | 9.43 | 7.11 | 3,077.02 | 3,077.02 | 12,940.58 | 100.0% |
| 128 | 512 | 2x | 53.01 | 18.74 | 4.11 | 4,068.08 | 4,750.31 | 13,319.07 | 100.0% |
| 128 | 512 | 4x | 112.84 | 38.63 | 2.29 | 3,718.67 | 6,181.44 | 13,144.89 | 100.0% |
| 128 | 512 | 8x | 242.34 | 78.24 | 1.22 | 2,956.45 | 5,256.42 | 13,112.60 | 100.0% |
| 128 | 512 | 16x | 439.37 | 155.77 | 0.72 | 3,818.20 | 8,686.77 | 13,221.55 | 100.0% |
| 128 | 512 | 32x | 808.63 | 307.13 | 0.42 | 3,928.24 | 7,642.88 | 13,322.80 | 100.0% |
| 128 | 512 | 64x | 1,707.19 | 601.86 | 0.24 | 3,750.13 | 7,458.30 | 13,479.67 | 100.0% |
| 128 | 512 | 128x | 3,187.28 | 1,191.72 | 0.15 | 4,158.98 | 8,525.86 | 13,579.68 | 100.0% |
| 128 | 512 | 256x | 6,142.45 | 2,267.84 | 0.09 | 4,280.16 | 8,383.73 | 14,241.42 | 100.0% |
| 128 | 512 | 512x | 10,677.49 | 4,090.53 | 0.06 | 4,829.05 | 9,994.34 | 15,223.58 | 99.6% |
| 128 | 512 | 1024x | 10,809.45 | 4,060.51 | 0.06 | 5,443.74 | 11,304.59 | 17,904.88 | 99.8% |
| 128 | 1,024 | 1x | 36.35 | 4.75 | 7.18 | 2,201.79 | 2,201.79 | 25,695.08 | 100.0% |
| 128 | 1,024 | 2x | 73.26 | 9.75 | 3.83 | 2,136.36 | 2,293.36 | 25,622.29 | 100.0% |
| 128 | 1,024 | 4x | 123.22 | 19.25 | 2.30 | 5,443.93 | 9,159.02 | 26,376.03 | 100.0% |
| 128 | 1,024 | 8x | 237.82 | 38.73 | 1.26 | 3,619.85 | 6,536.87 | 26,499.87 | 100.0% |
| 128 | 1,024 | 16x | 534.53 | 77.73 | 0.70 | 3,125.70 | 6,381.25 | 26,521.72 | 100.0% |
| 128 | 1,024 | 32x | 983.72 | 153.51 | 0.42 | 3,856.45 | 7,724.00 | 26,684.32 | 100.0% |
| 128 | 1,024 | 64x | 1,952.67 | 301.73 | 0.25 | 4,120.38 | 8,139.09 | 26,952.46 | 100.0% |
| 128 | 1,024 | 128x | 3,870.79 | 601.06 | 0.15 | 4,139.06 | 8,618.73 | 27,093.94 | 100.0% |
| 128 | 1,024 | 256x | 7,393.02 | 1,152.03 | 0.09 | 4,144.45 | 8,731.05 | 28,085.83 | 100.0% |
| 128 | 1,024 | 512x | 13,109.68 | 2,088.96 | 0.06 | 4,616.80 | 10,064.19 | 29,988.59 | 100.0% |
| 128 | 1,024 | 1024x | 13,387.14 | 2,113.53 | 0.05 | 5,436.52 | 12,043.20 | 34,013.07 | 100.0% |
| 128 | 2,048 | 1x | 37.28 | 2.41 | 7.37 | 4,010.73 | 4,010.73 | 50,586.54 | 100.0% |
| 128 | 2,048 | 2x | 68.77 | 4.80 | 4.03 | 2,748.56 | 3,062.12 | 51,777.39 | 100.0% |
| 128 | 2,048 | 4x | 134.44 | 9.87 | 2.27 | 4,301.92 | 7,133.85 | 51,483.21 | 100.0% |
| 128 | 2,048 | 8x | 257.55 | 19.33 | 1.29 | 3,158.81 | 4,844.99 | 53,120.02 | 100.0% |
| 128 | 2,048 | 16x | 551.39 | 38.66 | 0.72 | 3,067.25 | 4,870.34 | 53,325.36 | 100.0% |
| 128 | 2,048 | 32x | 973.88 | 75.73 | 0.45 | 3,704.26 | 8,225.97 | 54,121.08 | 100.0% |
| 128 | 2,048 | 64x | 2,070.60 | 150.22 | 0.26 | 4,172.48 | 9,610.59 | 54,204.33 | 100.0% |
| 128 | 2,048 | 128x | 4,070.38 | 298.51 | 0.16 | 3,830.38 | 7,843.86 | 54,691.97 | 100.0% |
| 128 | 2,048 | 256x | 6,340.65 | 475.44 | 0.11 | 4,549.79 | 10,167.82 | 68,610.58 | 100.0% |
| 128 | 2,048 | 512x | 11,752.38 | 891.82 | 0.07 | 4,908.31 | 10,435.49 | 72,379.55 | 100.0% |
| 128 | 2,048 | 1024x | 17,803.15 | 1,334.70 | 0.05 | 5,347.72 | 11,446.77 | 71,066.57 | 100.0% |
| 512 | 128 | 1x | 16.19 | 131.67 | 1.96 | 2,005.77 | 2,005.77 | 3,767.62 | 100.0% |
| 512 | 128 | 2x | 32.03 | 303.54 | 0.88 | 1,967.31 | 2,170.30 | 3,274.48 | 100.0% |
| 512 | 128 | 4x | 50.70 | 453.88 | 0.64 | 1,895.56 | 2,774.46 | 3,272.19 | 75.0% |
| 512 | 512 | 1x | 34.61 | 38.75 | 4.03 | 1,766.49 | 1,766.49 | 12,800.64 | 100.0% |
| 512 | 512 | 2x | 61.54 | 78.10 | 2.29 | 2,975.75 | 3,133.87 | 12,731.75 | 100.0% |
| 512 | 512 | 4x | 109.72 | 131.13 | 1.34 | 2,588.61 | 3,403.07 | 14,794.88 | 100.0% |
| 512 | 512 | 8x | 192.46 | 254.00 | 0.80 | 3,747.11 | 5,588.54 | 15,533.13 | 100.0% |
| 512 | 512 | 16x | 372.09 | 502.63 | 0.46 | 3,367.26 | 6,258.86 | 15,778.67 | 100.0% |
| 512 | 512 | 32x | 737.70 | 994.89 | 0.25 | 3,625.61 | 7,696.62 | 16,020.80 | 100.0% |
| 512 | 512 | 64x | 1,368.49 | 1,957.62 | 0.15 | 4,236.01 | 7,802.63 | 16,233.07 | 100.0% |
| 512 | 512 | 128x | 2,681.41 | 3,878.39 | 0.09 | 4,337.84 | 7,787.74 | 16,329.10 | 100.0% |
| 512 | 512 | 256x | 5,378.37 | 7,453.19 | 0.06 | 4,332.63 | 7,673.25 | 16,932.95 | 100.0% |
| 512 | 512 | 512x | 9,927.11 | 13,726.87 | 0.04 | 5,187.63 | 9,023.13 | 18,224.20 | 100.0% |
| 512 | 512 | 1024x | 12,940.84 | 17,877.99 | 0.03 | 6,822.27 | 11,975.33 | 20,922.65 | 100.0% |
| 512 | 1,024 | 1x | 37.26 | 19.31 | 5.22 | 1,702.51 | 1,702.51 | 25,680.21 | 100.0% |
| 512 | 1,024 | 2x | 69.56 | 38.07 | 2.96 | 2,940.78 | 3,953.92 | 26,110.94 | 100.0% |
| 512 | 1,024 | 4x | 142.88 | 75.11 | 1.60 | 2,079.75 | 3,173.83 | 26,324.82 | 100.0% |
| 512 | 1,024 | 8x | 266.26 | 148.87 | 0.89 | 3,586.80 | 4,818.39 | 26,529.32 | 100.0% |
| 512 | 1,024 | 16x | 498.32 | 296.25 | 0.53 | 2,641.61 | 3,766.36 | 26,768.24 | 100.0% |
| 512 | 1,024 | 32x | 957.37 | 590.10 | 0.31 | 3,795.71 | 7,229.33 | 27,022.98 | 100.0% |
| 512 | 1,024 | 64x | 1,887.02 | 1,161.15 | 0.18 | 4,183.92 | 7,757.14 | 27,406.28 | 100.0% |
| 512 | 1,024 | 128x | 3,714.66 | 2,287.07 | 0.11 | 3,705.88 | 6,989.96 | 27,717.40 | 100.0% |
| 512 | 1,024 | 256x | 7,292.38 | 4,391.45 | 0.07 | 4,088.79 | 7,593.77 | 28,818.02 | 100.0% |
| 512 | 1,024 | 512x | 13,667.35 | 8,185.37 | 0.05 | 4,648.08 | 8,559.81 | 30,679.13 | 100.0% |
| 512 | 1,024 | 1024x | 18,422.95 | 11,014.94 | 0.03 | 6,126.06 | 11,300.22 | 36,522.04 | 100.0% |
| 512 | 2,048 | 1x | 38.72 | 9.71 | 6.07 | 1,709.28 | 1,709.28 | 51,105.28 | 100.0% |
| 512 | 2,048 | 2x | 77.82 | 19.73 | 3.27 | 1,962.13 | 2,203.40 | 50,435.72 | 100.0% |
| 512 | 2,048 | 4x | 125.69 | 37.14 | 2.06 | 2,457.71 | 4,063.75 | 53,245.15 | 100.0% |
| 512 | 2,048 | 8x | 260.50 | 73.64 | 1.10 | 3,561.26 | 4,693.07 | 53,634.29 | 100.0% |
| 512 | 2,048 | 16x | 473.73 | 148.82 | 0.67 | 2,710.07 | 5,243.89 | 53,316.63 | 100.0% |
| 512 | 2,048 | 32x | 973.01 | 296.08 | 0.37 | 3,531.13 | 6,142.36 | 53,870.40 | 100.0% |
| 512 | 2,048 | 64x | 1,865.24 | 540.15 | 0.23 | 3,727.68 | 7,427.90 | 58,924.52 | 100.0% |
| 512 | 2,048 | 128x | 3,387.61 | 986.35 | 0.14 | 3,834.05 | 7,080.87 | 64,393.68 | 100.0% |
| 512 | 2,048 | 256x | 6,539.64 | 1,906.24 | 0.09 | 4,132.68 | 7,975.19 | 66,452.59 | 100.0% |
| 512 | 2,048 | 512x | 12,142.72 | 3,500.36 | 0.06 | 4,849.86 | 9,100.78 | 72,115.89 | 100.0% |
| 512 | 2,048 | 1024x | 16,881.70 | 4,859.38 | 0.04 | 7,134.87 | 13,065.30 | 74,691.45 | 100.0% |
| 1,024 | 512 | 1x | 31.90 | 74.40 | 2.74 | 2,459.80 | 2,459.80 | 13,103.45 | 100.0% |
| 1,024 | 512 | 2x | 62.83 | 148.09 | 1.50 | 2,524.26 | 2,982.04 | 13,182.09 | 100.0% |
| 1,024 | 512 | 4x | 106.22 | 289.96 | 0.83 | 4,069.58 | 5,583.48 | 13,475.73 | 100.0% |
| 1,024 | 512 | 8x | 211.45 | 571.13 | 0.46 | 4,021.75 | 6,156.26 | 13,710.42 | 100.0% |
| 1,024 | 512 | 16x | 424.65 | 1,144.37 | 0.27 | 3,984.15 | 5,791.10 | 13,719.03 | 100.0% |
| 1,024 | 512 | 32x | 807.81 | 2,267.30 | 0.15 | 4,156.02 | 7,436.66 | 13,828.89 | 100.0% |
| 1,024 | 512 | 64x | 1,591.82 | 4,488.01 | 0.09 | 4,610.53 | 8,769.62 | 13,975.35 | 100.0% |
| 1,024 | 512 | 128x | 3,044.34 | 8,569.54 | 0.06 | 4,553.99 | 7,904.51 | 14,498.94 | 99.2% |
| 1,024 | 512 | 256x | 5,530.89 | 15,918.78 | 0.04 | 5,492.89 | 9,795.60 | 15,625.76 | 99.6% |
| 1,024 | 512 | 512x | 10,024.74 | 28,435.84 | 0.03 | 6,587.42 | 10,515.66 | 17,514.71 | 100.0% |
| 1,024 | 512 | 1024x | 11,553.80 | 32,773.70 | 0.02 | 9,307.16 | 15,200.77 | 22,178.16 | 99.7% |
| 1,024 | 1,024 | 1x | 32.10 | 37.62 | 4.22 | 4,948.67 | 4,948.67 | 25,918.23 | 100.0% |
| 1,024 | 1,024 | 2x | 71.87 | 74.90 | 2.17 | 2,220.01 | 2,276.56 | 26,045.60 | 100.0% |
| 1,024 | 1,024 | 4x | 134.92 | 149.75 | 1.23 | 3,701.63 | 5,062.49 | 26,097.03 | 100.0% |
| 1,024 | 1,024 | 8x | 256.89 | 291.04 | 0.71 | 4,154.29 | 5,688.76 | 26,909.33 | 100.0% |
| 1,024 | 1,024 | 16x | 507.06 | 583.44 | 0.39 | 4,161.71 | 6,876.91 | 26,913.72 | 100.0% |
| 1,024 | 1,024 | 32x | 990.09 | 1,157.32 | 0.22 | 4,269.72 | 8,008.59 | 27,118.42 | 100.0% |
| 1,024 | 1,024 | 64x | 1,855.76 | 2,283.68 | 0.13 | 4,419.38 | 8,968.40 | 27,479.85 | 100.0% |
| 1,024 | 1,024 | 128x | 3,747.41 | 4,489.55 | 0.08 | 4,306.07 | 7,584.16 | 27,911.64 | 100.0% |
| 1,024 | 1,024 | 256x | 7,168.29 | 8,683.47 | 0.05 | 4,729.05 | 8,829.91 | 28,777.64 | 100.0% |
| 1,024 | 1,024 | 512x | 13,469.81 | 15,866.79 | 0.04 | 5,446.98 | 9,339.39 | 31,434.87 | 100.0% |
| 1,024 | 1,024 | 1024x | 14,863.28 | 17,687.89 | 0.03 | 8,036.43 | 13,298.05 | 38,284.75 | 100.0% |
| 1,024 | 2,048 | 1x | 35.34 | 18.71 | 5.39 | 5,322.01 | 5,322.01 | 52,099.10 | 100.0% |
| 1,024 | 2,048 | 2x | 72.16 | 37.76 | 2.90 | 4,578.91 | 7,086.77 | 51,699.53 | 100.0% |
| 1,024 | 2,048 | 4x | 138.51 | 72.72 | 1.66 | 4,777.69 | 6,566.19 | 53,743.72 | 100.0% |
| 1,024 | 2,048 | 8x | 286.94 | 147.61 | 0.86 | 3,355.72 | 4,878.05 | 53,054.09 | 100.0% |
| 1,024 | 2,048 | 16x | 525.35 | 290.55 | 0.52 | 4,461.39 | 9,856.01 | 54,063.35 | 100.0% |
| 1,024 | 2,048 | 32x | 1,082.02 | 581.19 | 0.29 | 4,019.11 | 6,924.52 | 54,034.35 | 100.0% |
| 1,024 | 2,048 | 64x | 2,018.03 | 1,143.42 | 0.18 | 3,951.20 | 7,211.16 | 54,968.16 | 100.0% |
| 1,024 | 2,048 | 128x | 3,970.46 | 2,266.20 | 0.11 | 4,459.57 | 8,807.74 | 55,423.77 | 100.0% |
| 1,024 | 2,048 | 256x | 7,699.61 | 4,402.35 | 0.07 | 4,934.34 | 9,009.11 | 56,988.94 | 100.0% |
| 1,024 | 2,048 | 512x | 14,166.95 | 8,019.96 | 0.05 | 6,108.63 | 10,214.85 | 62,440.23 | 100.0% |
| 1,024 | 2,048 | 1024x | 17,979.88 | 10,267.40 | 0.03 | 9,691.90 | 16,959.27 | 76,422.89 | 100.0% |
| 2,048 | 128 | 1x | 23.61 | 611.37 | 0.46 | 1,364.94 | 1,364.94 | 3,219.48 | 100.0% |
| 2,048 | 128 | 2x | 14.23 | 595.82 | 0.52 | 2,099.58 | 2,099.58 | 3,246.21 | 50.0% |
| 2,048 | 512 | 1x | 32.72 | 154.79 | 1.57 | 2,421.60 | 2,421.60 | 12,714.26 | 100.0% |
| 2,048 | 512 | 2x | 60.12 | 290.86 | 0.85 | 2,938.37 | 3,467.49 | 13,470.95 | 100.0% |
| 2,048 | 512 | 4x | 117.28 | 582.26 | 0.50 | 3,099.03 | 3,863.92 | 13,455.93 | 100.0% |
| 2,048 | 512 | 8x | 180.91 | 1,175.43 | 0.29 | 4,562.60 | 7,816.41 | 13,310.00 | 100.0% |
| 2,048 | 512 | 16x | 368.04 | 2,266.74 | 0.16 | 4,433.41 | 7,789.38 | 13,796.61 | 100.0% |
| 2,048 | 512 | 32x | 755.76 | 4,574.31 | 0.09 | 4,516.34 | 7,800.66 | 13,647.61 | 100.0% |
| 2,048 | 512 | 64x | 1,457.47 | 8,613.28 | 0.06 | 5,041.78 | 8,266.29 | 14,480.79 | 100.0% |
| 2,048 | 512 | 128x | 2,769.19 | 16,120.56 | 0.04 | 5,645.28 | 9,260.51 | 15,452.64 | 100.0% |
| 2,048 | 512 | 256x | 5,021.59 | 29,872.90 | 0.03 | 6,876.39 | 11,113.59 | 16,646.74 | 100.0% |
| 2,048 | 512 | 512x | 8,418.47 | 50,200.55 | 0.02 | 9,143.15 | 14,351.75 | 19,613.36 | 99.6% |
| 2,048 | 512 | 1024x | 8,273.31 | 48,936.11 | 0.02 | 14,204.02 | 21,115.78 | 27,235.20 | 99.8% |
| 2,048 | 1,024 | 1x | 32.98 | 75.81 | 2.70 | 4,212.75 | 4,212.75 | 25,958.98 | 100.0% |
| 2,048 | 1,024 | 2x | 65.15 | 149.84 | 1.48 | 4,425.84 | 4,497.76 | 26,137.27 | 100.0% |
| 2,048 | 1,024 | 4x | 139.40 | 297.99 | 0.75 | 2,894.42 | 3,410.41 | 26,302.99 | 100.0% |
| 2,048 | 1,024 | 8x | 200.65 | 582.33 | 0.49 | 3,372.38 | 5,789.74 | 26,862.81 | 100.0% |
| 2,048 | 1,024 | 16x | 416.84 | 1,157.91 | 0.27 | 4,355.46 | 8,889.66 | 26,998.09 | 100.0% |
| 2,048 | 1,024 | 32x | 864.25 | 2,305.10 | 0.15 | 4,424.43 | 7,512.36 | 27,104.86 | 100.0% |
| 2,048 | 1,024 | 64x | 1,772.18 | 4,513.52 | 0.09 | 4,540.30 | 8,174.91 | 27,677.38 | 100.0% |
| 2,048 | 1,024 | 128x | 3,579.53 | 8,716.48 | 0.06 | 4,997.28 | 9,111.13 | 28,644.04 | 100.0% |
| 2,048 | 1,024 | 256x | 6,737.22 | 16,657.76 | 0.04 | 6,006.98 | 10,286.20 | 29,925.44 | 100.0% |
| 2,048 | 1,024 | 512x | 12,109.00 | 29,635.29 | 0.03 | 8,028.69 | 12,449.47 | 33,561.80 | 100.0% |
| 2,048 | 1,024 | 1024x | 15,343.33 | 37,646.45 | 0.02 | 14,487.42 | 20,218.99 | 45,000.76 | 100.0% |
| 2,048 | 2,048 | 1x | 38.49 | 38.86 | 3.80 | 2,481.14 | 2,481.14 | 50,636.77 | 100.0% |
| 2,048 | 2,048 | 2x | 74.67 | 75.40 | 1.97 | 2,784.15 | 2,795.01 | 51,974.62 | 100.0% |
| 2,048 | 2,048 | 4x | 104.48 | 151.64 | 1.26 | 3,656.24 | 4,795.75 | 51,600.04 | 100.0% |
| 2,048 | 2,048 | 8x | 192.12 | 292.94 | 0.76 | 3,973.53 | 6,003.97 | 53,418.88 | 100.0% |
| 2,048 | 2,048 | 16x | 372.27 | 583.07 | 0.43 | 4,096.18 | 7,922.12 | 53,626.70 | 100.0% |
| 2,048 | 2,048 | 32x | 874.58 | 1,157.18 | 0.23 | 4,369.15 | 7,449.95 | 54,042.98 | 100.0% |
| 2,048 | 2,048 | 64x | 1,825.17 | 2,263.66 | 0.14 | 5,223.31 | 9,970.30 | 55,233.80 | 100.0% |
| 2,048 | 2,048 | 128x | 3,846.41 | 4,423.24 | 0.08 | 5,591.07 | 9,451.80 | 56,494.14 | 100.0% |
| 2,048 | 2,048 | 256x | 7,204.48 | 8,512.34 | 0.05 | 6,826.63 | 11,870.12 | 58,669.83 | 100.0% |
| 2,048 | 2,048 | 512x | 13,178.61 | 15,408.00 | 0.04 | 9,057.79 | 12,811.69 | 64,681.28 | 100.0% |
Runtime parameters used across all benchmark runs
