Inference

Speculative decoding has moved from inference-paper trivia into the local LLM hot path. The useful version now includes ngram speculation, native MTP heads, DFlash-style drafting, and TurboQuant / TCQ KV compression, with early llama.cpp forks showing both real speedups and real rough edges.

Deep dive into the differences between Z-Image (the full foundation model) and Z-Image Turbo (the distilled 8-step variant). Includes technical specs, community benchmarks, VRAM requirements, and practical usage recommendations.