Sign in Subscribe

Topic

AI Benchmarking

A collection of 2 issues

Enterprise AI Doesn't Need Enterprise Hardware

Four obsolete GPUs worth $12,200 delivered performance for sovereign AI. 8x better throughput with SGLang vs Ollama.

Prem AI Adds DeepSeek-V3.1 for Smarter Enterprise AI

PremAI now supports DeepSeek-V3.1, a hybrid MoE model with 128K context, smart routing, and benchmark gains, built for enterprise use with secure, ready-to-deploy APIs.