Last released Oct 15, 2026
Deterministic high-precision VRAM, GQA KV-Cache, MoE Memory Bandwidth, and llama.cpp layer offloading profiler for 2026 AI models and silicon.