List your company

Decode Throughput Generation on NVIDIA B200 GPU

6,742.39Throughput (tokens/s)YOCO (CLSA)
178.71081,882.74293,586.7755,290.8071Jun 4, 2026

Evaluation results

MethodLinks
YOCO (CLSA)Context Length=8K2026.066,742.39
YOCO (CLSA)Context Length=16K2026.066,350.39
YOCO (Dense)Context Length=8K2026.065,516.5
YOCO (CLSA)Context Length=32K2026.065,461.33
TransformerContext Length=8K2026.064,762.79
YOCO (Dense)Context Length=16K2026.064,147.85
YOCO (CLSA)Context Length=64K2026.064,137.37
YOCO (CLSA)Context Length=128K2026.063,276.8
TransformerContext Length=16K2026.063,091.32
YOCO (Dense)Context Length=32K2026.062,677.12
TransformerContext Length=32K2026.061,761.72
YOCO (Dense)Context Length=64K2026.061,600
YOCO (Dense)Context Length=128K2026.06960.94
TransformerContext Length=64K2026.06948.15
TransformerContext Length=128K2026.06431.16