Qwen3.6 9B will release around Google I/O?
Posted by BothYou243@reddit | LocalLLaMA | View on Reddit | 20 comments
I don't think alibaba officially stated about "no qwen3.6 smaller models", and according to the patterns, she should ave been released it in the first week of may, but I think they delayed a little bit to catch the spotlight from Google I/O, as maybe google is preping to advertise more about new gemma?
what do you guys think?
and I am eagerly waiting for qwen3.6 9B, as I don't think qwen3.6 27B at lower quant (2bit) is good for mac mini m4 (16GB variant)
VoiceApprehensive893@reddit
gemma 4 14b , audio input for all gemma 4 models and >100b MoE trust
DertekAn@reddit
Gemma 4 14B? Really?
Long_comment_san@reddit
Would be quite a banger
BeautyxArt@reddit
how fare qwen27b slower than qwen9b with same hw ?
ea_man@reddit
3x ?
EbbNorth7735@reddit
I think they've moved on from 3.6. If we're lucky we'll get some 3.7 models in a month or so.
BeautyxArt@reddit
3.7 petabyte HDDs needed.
JGeek00@reddit
Is there a public roadmap for new versions?
my_name_isnt_clever@reddit
Nope. Everything people say here about Qwen is based on vibes and a tweet or two.
DeltaSqueezer@reddit
Could be. They already released the 2 popular sizes. I'd hope they'd release the smaller ones too, but it could well be that with limited compute, they might focus on the next generation instead.
VoiceApprehensive893@reddit
gemma 4 14b and >100b MoE trust
cartazio@reddit
i want more base models released. 3.5 9b base is wonderfully honest
NNN_Throwaway2@reddit
Currently all evidence points to no more Qwen 3.6. Enjoy 3.5, it id still very good.
dampflokfreund@reddit
Hope for fixed Gemma 4.1 QAT with audio input for all models, better performance and better tool calling.
stoppableDissolution@reddit
Also the supposedly-unreleased 120b+ wouldve been so nice to have. I really hoped for mistral to deliver on that front but it was kinda disappointing
Alternative_You3585@reddit
The release will drown under the Google stuff, I doubt there will be any releases except Google next week from smaller LLM companies
BothYou243@reddit (OP)
Alibaba? smaller llm company?
Alternative_You3585@reddit
Compared to openai, deepmind or anthropic yes. While sure release make a lot of noise and got a lot of local users, it ain't the same as 5T anthropic models (leaked by musk) or assumably even bigger by Google
SnooPaintings8639@reddit
It's like HL3 all over again.
sn2006gy@reddit
Once can only hope..
I'd love to see Qwen recover from their mass layoff that seemed to insinuate a dwindling effort on open weight models.