Running Qwen3.8-Flash-Next NVFP4 at 184 tok/s on a single RTX 6000 Pro

27 points · 4 visible top comments · 2026-10-26 22:17:22 UTC

Comments

lowbloodsugar · 2026-10-27 00:58:22 UTC

Been running Qwen3.8-Flash-Next NVFP4 for weeks now. What's new about this?

catchnear4321 · 2026-10-27 01:20:43 UTC

hn title is completely different from the article title, which takes clicking the hn title to discover.

> Our first model-kernel combos

cjbprime · 2026-10-27 18:37:11 UTC

They did a specific-hardware kernel optimization for the Pro 6000 (available on AWS as g7e).

IronWolve · 2026-10-27 02:59:02 UTC

Been running versions of this on my pc, if you have 64 gigs ram, you can run it on a 12+ gig cards easily.

saejox · 2026-10-27 12:28:32 UTC

it is very hard to run 150k token context with these kind of setups. even if you do, all of us are used to running multiple agents at once.

i hope ram prices will come down a little.

aappleby · 2026-10-27 18:43:39 UTC

On "just" a RTX 6000 Pro? You mean one of those $16,000 gpus?