Qwen 3.8-Flash-Next releasing tomorrow (125B a6B)

#1

Post by garo-pro » Tue, Aug 25, 2026, 11:49 AM UTC

Re: Qwen 3.8-Flash-Next releasing tomorrow (125B a6B)

#2

Post by tarruda » Tue, Aug 25, 2026, 12:00 PM UTC

Can you share the source for the parameter count (125B A6B)? I didn't see it anywhere in the page.

#3

Post by petu » Tue, Aug 25, 2026, 12:37 PM UTC

tarruda wrote: post #2

Can you share the source for the parameter count (125B A6B)? I didn't see it anywhere in the page.

It was in description under the countdown initially, but was quickly removed.

It also said 51B of n-grams and new attention (IIRC it said "Qwen Sparse Attention").

#4

Post by fcanesin » Tue, Aug 25, 2026, 1:25 PM UTC

#5

Post by honestlyranked » Tue, Aug 25, 2026, 1:26 PM UTC

Alibaba is giving sleepless nights to the tech giants

#6

Post by mrdoe » Tue, Aug 25, 2026, 1:29 PM UTC

lol blocked with dns4eu

what a joke this resolver has become

#7

Post by cogman10 » Tue, Aug 25, 2026, 1:29 PM UTC

Wow. I wasn't expecting this. I thought they were going to do a 35B model instead.

#8

Post by big-chungus4 » Tue, Aug 25, 2026, 1:33 PM UTC

I hope there is going to be a free endpoint... Unlike 35B-A3B, I am nowhere close to running it locally

#9

Post by blurbleblurble » Tue, Aug 25, 2026, 1:34 PM UTC

gg

#10

Post by big-chungus4 » Tue, Aug 25, 2026, 1:38 PM UTC

> We are releasing these architectural improvements ahead of time so that the community can prepare for the upcoming full family of Qwen4 models.

That gives me hope that "full family" means it will include smaller models like 4B.