@0xSero: Best models smallest to largest right now. - Gemma-4-12B - Qwen3.8-27B - Laguna-S2.1 - Deepseek-V4-Flash - Inkling-Sm…
0xSero
@0xSero
Best models smallest to largest right now.
- Gemma-4-12B
- Qwen3.8-27B
- Laguna-S2.1
- Deepseek-V4-Flash
- Inkling-Small
- MiniMax-M3
- GLM-5.*
- Kimi-K2.7-Code
- Qwen3.8-Max
- Kimi-K3
Spoiled for choice, open weights community is much more exciting than the frontier.
💬 135 · 🔁 182 · ❤️ 2.4K · 👁 104.3K
11:02 PM · Aug 3, 2026 · View on X
Replies (22)
Keven Crick @keven_ink
Why are you ranking qwen 3.8 27:b when it's not out yet? unless you have some kind of early access, you're better than this...
0xSero @0xSero
You must only know where a thing has been to know where it’s going.
Nishaant @Insigolon
What's your opinion on glm 4.7 flash been lookin to switch to it as a daily driver. Doesn't have vision
0xSero @0xSero
qwen is better
Black Eyed Hat Man @BlackEyedHatMan
With the way you're carrying on...
Soon you'll need to have a word with Copenhagen Atomics! 😉
* Superb to see, you're inspiring us all 🙌
invalidbug @invalidbug
Forget cost, how are you powering that without tripping a breaker?
AYO MAYO @AYOMAYO180427
That cpu cooler is not correctly mounted it should be sideways just keep that in mind if you push it while pushing the gpus also
I understand that it is a space constraint but perhaps a smaller cooler would be better than that one incorrectly mounted
CodeBuilder_ @CodeBuilder_
Laguna S2.1 or Inkling Small?
Kim Aurélie Moreau @kim__aurelie
is that a quad RTX 6000 Pro Stack? 😱
Miloš van DotNET ✡︎ @msdevelx
Jesus. Great PC and setup and everything. But those fans are terrible. The black one from noctua would be much better.
NSD @NSD764464711907
My only problem with Kimi k3 is it hallucinates being claude just too much.
killian @grisetio
what about smaller than 12b ?
not for coding but prose
Hudson Gouge @HudsonGouge
Qwen3.8-27B isn't even available yet.
Sweetest Evana @SweetestEvana
Hmmm nah laguna and Ornith are flavours of Qwen… I don’t think they ever perform as well…
Steven Cheng @stevencheng
My GPU fans are crying just reading this.
aphex @aphexinvests
HM gemma e4b, punches way above weight especially with search
Ryan @RyanAtlasReyes
Im using deepseek v4 flash
With pi (inspiration from you)
Does your model ever send incomplete responses like this?
Look at the time stamps in replies for reference.
Also my conversation is about a chess game.
John T Davies @jtdavies
I think the Laguna XS 2.1 should be in there, it's better than Gemma-4, not quite as good as Qwen3.6. I think you meant Qwen3.6 as 3.8 isn't out yet.
Cody Schuldt @codylschuldt
Qwen 3.8-27b isn't out yet...
dougvk @dougvk
qwen 3.8 27b isnt out yet?
HB @HoneyBadgar
What models do you like running on the AI 395 Max chip
Coconut @faizrazz
Were you given early access to 3.8 27 B?
Comments (0)
No comments yet