Ling-3.0-flash MXFP4 MoE GGUF released with MTP, plus a heretic variant

pmttyji · reddit · 2026-08-18

Community quantizer noctrex released MXFP4 MoE GGUF versions of Ling-3.0-flash on Hugging Face, with built-in MTP (multi-token prediction) support for local inference.

A "heretic" variant with loosened guardrails is also available.

Original post →

More from Models

Models channel →