List your company

LLM Serving and Question Answering on MultiHopRAG hot0.7 (test)

49.27KV-Cache Hit RatePRISM
8.408419.016729.62540.2333May 9, 2026

Evaluation results

MethodLinks
PRISMModel=Qwen3-4B-Instruct-2507, QPS=802026.0549.2740.28
PRISMModel=Qwen3-4B-Instruct-2507, QPS=602026.0549.2540.29
PRISMModel=Qwen3-4B-Instruct-2507, QPS=702026.0549.140.44
PRISMModel=Qwen3-4B-Instruct-2507, QPS=402026.0548.9840.61
PRISMModel=Qwen3-4B-Instruct-2507, QPS=502026.0548.940.33
ContextPilotModel=Qwen3-4B-Instruct-2507, QPS=602026.0543.7639.36
ContextPilotModel=Qwen3-4B-Instruct-2507, QPS=402026.0543.6440.33
ContextPilotModel=Qwen3-4B-Instruct-2507, QPS=502026.0543.5139.85
ContextPilotModel=Qwen3-4B-Instruct-2507, QPS=802026.0543.2440.56
ContextPilotModel=Qwen3-4B-Instruct-2507, QPS=702026.0542.0439.48
PRISMModel=Llama2-13B, QPS=302026.0540.1521.88
PRISMModel=Llama2-13B, QPS=102026.0539.7921.69
PRISMModel=Llama2-13B, QPS=252026.0539.2221.79
PRISMModel=Llama2-13B, QPS=202026.0539.1521.36
PRISMModel=Llama2-13B, QPS=152026.0539.0221.58
ContextPilotModel=Llama2-13B, QPS=252026.0528.1420.67
ContextPilotModel=Llama2-13B, QPS=302026.0527.520.58
ContextPilotModel=Llama2-13B, QPS=202026.0527.0620.48
ContextPilotModel=Llama2-13B, QPS=152026.0527.0120.98
ContextPilotModel=Llama2-13B, QPS=102026.0526.5120.97
k-LPMModel=Llama2-13B, QPS=302026.0520.8620.66
k-LPMModel=Qwen3-4B-Instruct-2507, QPS=702026.0520.3939.76
k-LPMModel=Llama2-13B, QPS=202026.0520.3420.56
k-LPMModel=Qwen3-4B-Instruct-2507, QPS=602026.0520.2639.76
k-LPMModel=Llama2-13B, QPS=152026.0520.1420.45
k-LPMModel=Llama2-13B, QPS=102026.0520.121.01
k-LPMModel=Qwen3-4B-Instruct-2507, QPS=402026.0519.9439.81
k-LPMModel=Llama2-13B, QPS=252026.0519.8920.68
k-LPMModel=Qwen3-4B-Instruct-2507, QPS=802026.0519.8639.81
k-LPMModel=Qwen3-4B-Instruct-2507, QPS=502026.0516.9339.81
native SGLangModel=Qwen3-4B-Instruct-2507, QPS=802026.0516.0639.48
native SGLangModel=Qwen3-4B-Instruct-2507, QPS=402026.0515.1739.76
native SGLangModel=Qwen3-4B-Instruct-2507, QPS=602026.0515.1439.81
native SGLangModel=Qwen3-4B-Instruct-2507, QPS=502026.0515.0739.81
native SGLangModel=Qwen3-4B-Instruct-2507, QPS=702026.0514.9839.53
native SGLangModel=Llama2-13B, QPS=302026.0511.0620.98
native SGLangModel=Llama2-13B, QPS=102026.0510.1720.78
native SGLangModel=Llama2-13B, QPS=202026.0510.1420.74
native SGLangModel=Llama2-13B, QPS=152026.0510.0720.88
native SGLangModel=Llama2-13B, QPS=252026.059.9820.96