I hope it's K3 level and below 1T, we need more competition!
Edit: GLM dropped
GLM-5.3 dropped today! (14/8/2026)
open weights release in ~2 weeks!
Published benchmarks:
GLM-5.3 Imminent drop
https://github.com/zai-org/z-ai-sdk-java/commits/glm-5.3
https://github.com/zai-org/z-ai-sdk-java/commits/glm-5.3
I hope it's K3 level and below 1T, we need more competition!
GLM-5.3 dropped today! (14/8/2026)
open weights release in ~2 weeks!
Published benchmarks:
11 Comments
SirDimples@programming.dev · 4 pts · 32d
Appears they removed the branch and changesets
leanleft@lemmy.ml · 4 pts · 32d
glm 5.2 is already pretty good, soo...
librekitty@lemmy.today · 1 pts · 31d
quite large though!
leanleft@lemmy.ml · 2 pts · 30d
on the extreme opposite side of the spectrum:
https://huggingface.co/LiquidAI/LFM2.5-2.6B
or.. really extreme.. https://huggingface.co/AxiomicLabs/GPT-X2.5-135M
Jrockwar@feddit.uk · 2 pts · 29d
Are these any good? I mean, within their size class, obviously - not expecting them to compare to Kimi K3!
Models the size of that 135M one open up some interesting use cases for edge devices.
humanspiral@lemmy.ca · 2 pts · 21d
it's by far best sub trillion parameter model. fits in 512gb with 1m context at q4. Or less rental hardware than any model that scores higher than it.
BeefAndPoultry@lemmus.org · 2 pts · 32d
this has been a crazy few weeks! lol
cyd@lemmy.world · 2 pts · 32d
Very likely the same architecture, just with more post training.
TheLegendaryAssholeOfJushinLiger@sh.itjust.works · 0 pts · 32d
What makes you say that? Not arguing, just curious.
cyd@lemmy.world · 4 pts · 31d
Just repeating rumors (sorry, should have been clearer): GLM 5.3 is 5.2 with extra post training, their big upcoming model is 5.5. It kind of makes sense too; pretraining on a new architecture/size is expensive and it's natural for these companies to try to wring an extra minor version or two out of each one.
SirDimples@programming.dev · 1 pts · 21d
Update: GLM-5.3 dropped today, open weights release in ~2 weeks!