The short version
A vLLM fork adding MXFP4 and early PARO quantization for AMD R9700 GPUs.
From the HackerLinks archive
Radiance vLLM MXFP4
A vLLM fork adding MXFP4 and early PARO quantization for AMD R9700 GPUs.
At a glance:
First seen:2026-09-08
Last seen:2026-09-08
Times seen:1
Website:codeberg.org
Why it caught our attention
An AMD GPU user called the fork excellent and uses it daily.
Where it surfaced on Hacker News
2026-09-08
“The MXFP4 fork is excellent too. Its my daily driver right now.”
intothemild · recommendation · first hand use
Direct HN comment
Speculative Decoding in vLLM on AMD GPUs