SemiAnalysis launches InferenceMAX, an open-source benchmark that automatically tracks LLM inference performance across AI models and frameworks every night (SemiAnalysis)

SemiAnalysis:
SemiAnalysis launches InferenceMAX, an open-source benchmark that automatically tracks LLM inference performance across AI models and frameworks every night  —  NVIDIA GB200 NVL72, AMD MI355X, Throughput Token per GPU, Latency Tok/s/user, Perf per Dollar, Cost per Million Tokens …

from Techmeme https://ift.tt/Xh43uE0