Hacker News new | past | comments | ask | show | jobs | submit
pettijohn | on: Qwen3.8-Max: A New Bar for Coding and Cowork
35B MoE is certainly a good and fast local model. I find 27B dense to be quite a bit smarter, so I daily drive that. I wish there was a ~100B MoE with maybe 10B active. It would be super smart and fast!
mattnewton | parent | next
There was a 3.5 122B 10A release -

https://huggingface.co/Qwen/Qwen3.5-122B-A10B

kanemcgrath | root | parent
I tried it for a bit, and It was not really worth its size. It got swept up in all the other AI news recently, but laguna s 2.1 I think is the best ~100B moe model right now
nozzlegear | root | parent | next
I didn't mention it above, but Laguna S is my other favorite model. I use Qwen a lot more, it's smaller and faster, but I like to switch to Laguna when I feel like I need a "heavy hitter" for certain huge or complex tasks.
tommica | root | parent | next
What on earth hardwares do you guys have to be able to run 100gb models locally?! That's crazy! I'm here struggling to even get 27b models to run in somewhat usable way
loading story #49155871
loading story #49151535
numpad0 | root | parent | next
Yeah, here I am sitting deeply deeply deeply regretting not buying couple CMP 170HX at $200 or $350, knowing I could just flip them ethically at purchase price if nothing came of it... I could have just casually built a 128GB dual A100 local AI monster
loading story #49154585
loading story #49152339
sarjann | root | parent | next
dgx spark, nvfp4 so I have spare room for KV cache (context)
loading story #49151994
loading story #49155401
loading story #49151547
antman | parent | next
A 35 A3B as smart as previous gen 27B would be a sweet point
traceroute66 | parent | next
> 35B MoE is certainly a good and fast local model. I find 27B dense to be quite a bit smarter

Isn't that just the definition of MoE vs dense ?

cyanydeez | root | parent
Full name is 35B-A3B. 3B Is the token generator thats selected out of the 35b available in the model, which is some layered jazz.

So it can be dumber but its quite capablr.

wickedsight | parent | next
I use 27B in plan mode and 35B MoE in act mode. I noticed that is the best balance for me for consistent tool calls and intelligent planning. Takes some time to switch, but it's worth it for me.
julianlam | root | parent
Wonder if it's possible to share a common cache via cachyllama between 27B and 35B-A3B
nozzlegear | parent
I've heard 27B is smarter! I tried it some time ago but couldn't get it working with my oMLX. I need to try it again.
mattnewton | root | parent | next
Honestly the 27b dense one punches way above its weight in a lot of domains, especially coding in my testing, so I think you will probably be disappointed.
npodbielski | root | parent
In my case I would say they are comparable but moe models are looping and getting lost a lot more than dense models.

On the other hand having 90t/s with any local model is nice and Pi with loop police extension can prevent looping a lot.

loading story #49156021