List your company

Prefill Throughput on 128K context

7,756.8Prefill Throughput (tok/s)SparDA
4,143.9445,081.8976,019.856,957.803Jun 3, 2026

Evaluation results

MethodLinks
SparDABackbone=MiniCPM4.1-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.067,756.8
SparseBackbone=MiniCPM4.1-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=false2026.066,310.3
SparseBackbone=MiniCPM4.1-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.066,299.2
InfiniGenBackbone=MiniCPM4.1-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.066,296.3
SparDABackbone=NOSA-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.065,477.5
SparseBackbone=NOSA-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.064,717.1
SparseBackbone=NOSA-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=false2026.064,713.5
InfiniGenBackbone=NOSA-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=true2026.064,711.5
DenseBackbone=NOSA-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=false2026.064,316
DenseBackbone=MiniCPM4.1-8B, Hardware=NVIDIA A100 GPU, Batch Size=4, KV-offloading=false2026.064,282.9