this post was submitted on 10 Oct 2026

27 points (86.5% liked)

LocalLLaMA

localllama@sh.itjust.works

5,331 readers

616 active

23 users here now

Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.

Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.

As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.

Rules:

Rule 1 - No harassment or personal character attacks of community members. I.E no namecalling, no generalizing entire groups of people that make up our community, no baseless personal insults.

Rule 2 - No comparing artificial intelligence/machine learning models to cryptocurrency. I.E no comparing the usefulness of models to that of NFTs, no comparing the resource usage required to train a model is anything close to maintaining a blockchain/ mining for crypto, no implying its just a fad/bubble that will leave people with nothing of value when it burst.

Rule 3 - No comparing artificial intelligence/machine learning to simple text prediction algorithms. I.E statements such as "llms are basically just simple text predictions like what your phone keyboard autocorrect uses, and they're still using the same algorithms since <over 10 years ago>.

Rule 4 - No implying that models are devoid of purpose or potential for enriching peoples lives.

founded 3 years ago

moderators

MonsterBug@sh.itjust.works

Smokeydope@lemmy.world

pax@sh.itjust.works

noneabove1182@sh.itjust.works

What Qwen4 sizes are you hoping for? (lemmy.ml)

submitted 1 week ago by robber@lemmy.ml to c/localllama@sh.itjust.works

18 comments fedilink hide all child comments

Basically the title.

For me, a dense model around 20b, with an additional engram table of 10-20b, or a sparse MoE around 80b would be amazing.

all 18 comments

sorted by: hot top controversial new old

[–] swelter_spark@reddthat.com 1 point 6 days ago

30-35B or just under seems perfect to me.

[–] fancy_pants@reddthat.com 1 point 6 days ago

Something that would make good use of a 128GB spark. GPT-OSS-120b is good but shows its age

[–] e0qdk@reddthat.com 12 points 1 week ago (1 child)

Another 35B-A3B MoE or similar would be nice...

FYI, Yandex released an 80B MoE base model (i.e. not instruction tuned!) recently: https://huggingface.co/yandex/AliceAI-Foundation-80B-A3B-Base -- haven't experimented with it personally though.

hideshow 1 child comment

[–] robber@lemmy.ml [S] 2 points 1 week ago

I hope an instruction tuned will be released as well.

parent

[–] Eyekaytee@aussie.zone 12 points 1 week ago (4 children)

gemma 26b-a4b was fantastic, anything in that area is ideal

running my tiny rx6600 8gb vram

hideshow 4 child comments

[–] hoshikarakitaridia@lemmy.world 3 points 1 week ago* (2 children)

Oh hell yeah

I'm on my 1070 8gb copium so we're twinsies

Maybe they release a huge moe model with non-uniform quants and then we can get even more quality on those cute little cards

parent

hideshow 2 child comments

[–] AceBonobo@lemmy.world 1 point 1 day ago (1 child)

How's your experience with the 1070?

parent

hideshow 1 child comment

[–] hoshikarakitaridia@lemmy.world 1 point 1 day ago

Mixed?

It's the lower bound of where LLMs are helpful. Anything less and the models would be either too slow to work or too dumb. With the 1070 they are already painfully slow but at least they're still good enough to do some menial tasks.

A good harness for web search etc does a lot though.

parent

[–] ComradePedro@lemmy.ml 1 point 1 week ago

I have an M4 Pro 24GB and I still use Gemma 26B-A4B as my primary model. Beats every other model I’ve ever tried locally on most stuff except coding and tool use, so a Qwen 4 at this size would be super interesting

parent

[–] dosse91@lemmy.trippy.pizza 10 points 1 week ago (1 child)

Flash next 125b is genuinely amazing, I hope to see 4.0 of the same size.

hideshow 1 child comment

[–] tpWinthropeIII@lemmy.world 2 points 1 week ago

Agree, and upvoted.

Another good size would be 160 gb for 96 gb vram systems. After that I would like to see 225 to 250 gb.

parent

[–] BeefAndPoultry@lemmus.org 5 points 1 week ago

32GB RAM and 8GB VRAM here, so 35b a3b is about my maximum, maybe with added n-grams

[–] fhein@lemmy.world 4 points 1 week ago (3 children)

27B dense was a good size for 3.8, but I'm guessing they're going to add a few B parameters in order to make it noticeably better.

hideshow 3 child comments

[–] BeefAndPoultry@lemmus.org 1 point 1 week ago*

If they can make the KV cache size scale more efficiently with many tokens, like DeepSeek 4.1 Flash fitting 1 million tokens in 890MB, it might give them enough room to add a couple billion parameters of weights and still fit well in 32GB GPUs

parent

[–] robber@lemmy.ml [S] 1 point 1 week ago (1 child)

Not necessarly, I'd say. We've seen many releases with multiple sizes and the 27B class has been around since at least the Gemma3 times.

parent

hideshow 1 child comment

[–] fhein@lemmy.world 3 points 1 week ago

Well, it's just a guess. But also Gemma4 increased to 31B, so I think there's a possibility that Qwen does the same.

parent

[–] ZephyrXero@lemmy.world 2 points 1 week ago (1 child)

They've changed the license with Qwen 4, starting with 3.8-next. So I probably won't be using any further releases

hideshow 1 child comment

[–] fluxx@lemmy.world 4 points 1 week ago

What are the significant changes? Requiring attribution or something more sinister?

parent