DeepSeek V4.1 Flash on GB10: 67 to 140.9 tok/s scaling 1 to 8 chats

On the same GB10 boxes and engine, DeepSeek V4.1 Flash goes from 67 to 140.9 tok/s scaling one to eight chats, versus GLM 5.3 Flash at 88.1 to 177.2.

CryptoBandit

@CryptoBandit100Archived 9 days agoOriginally posted 2026.10.23

CryptoBandit @CryptoBandit100Originally posted 2026.10.23

Open original on X

Same 4 GB10 boxes, same engine, scaling from 1 to 8 chats: GLM 5.3 Flash 88.1 to 177.2 tok/s DeepSeek V4.1 Flash 67 to 140.9 tok/s Temp 0, thinking off, 512 tokens, median of 3. GLM adds more per chat too. Do you size a home rig for one chat or for many?

x.com/cryptobandit100/status/2113389938834366947Archived 2026.10.23 12:05

Open original on X