Benchmark Leaderboard

Official and community-submitted performance benchmarks for KTransformers

Submit Benchmark
Filter by:
#PrecisionGPUCPUVersion
1DeepSeek-R1-0528 671B
FP8
8x L202x Intel Xeon 6454S227.8587.58
0.4
2MiniMax-M2.1
FP8
2x RTX 50902x AMD EPYC 93554007.0033.10
0.4
3MiniMax-M2.1
FP8
1x RTX 50902x AMD EPYC 9355408.0032.10
0.4
4MiniMax-M2.1
FP8
1x RTX 50902x AMD EPYC 93551196.0031.40
0.4
5MiniMax-M2.1
FP8
1x RTX 50902x AMD EPYC 93552540.0027.60
0.4
6MiniMax-M2.1
FP8
2x RTX 40902x Intel Xeon 8488C2269.0021.60
0.4
7MiniMax-M2.1
FP8
1x RTX 40902x Intel Xeon 8488C1385.0018.50
0.4
8DeepSeek-R1/V3 671B
BF16
1x RTX 4090D2x Intel Xeon 6454S286.5514.20
0.3
9DeepSeek-R1/V3 671B
Q4_K_M
1x RTX 4090D2x Intel Xeon 6454S97.3213.69
0.2
10DeepSeek-V3 671B
Q4_K_M
2x Intel Xeon 6454S10.314.51
Showing 10 of 10 benchmarks