New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Looking to trade, Succulent, KS-LE-B or MYSTERY boxes
Neoon
Community Contributor, Veteran
Hi,
Offerings
https://stats.neoon.net/monitor/639326985a0f1c8929a95192727bebd2
https://stats.neoon.net/monitor/b529c94af9f89e6235351a7891a985c2
TLDR: KS-LE-B in GRA and BHS.
Objective
I need more RAM for LLM's, even its slow, I want to build at least a 128gig cluster.
Means I need at least 2x KS-LE-B in one DC or a fat 128gig Mystery.
I don't want to pay shiton of money for a 128gig or bigger box.
I am looking to trade so, I can archive this goal.
Anyone of you willing to trade?
Looking 1:1 but I am willing to accept less storage.
If you are willing to just sell me one, either in GRA or BHS.
Yes but no, if you desire $$$, no, $$ could be doable.
Anyway, thanks by. luv yo.
Comments
Available for outright sale?
Just out of curiosity, what LLMs are you running on your servers?
Glorious leader, gib BHS server plz
DM‘d
A lot, I use 1.2TB+ for the models alone.
I'm also curious what model is running. I have a 128GB Hetzner dedi becoming idle soon. I don't want to release it or transfer it.
Right now.
https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF
https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF
Playing with these, since they a new.
But also GTP OSS 120b, GLM Flash 4.7, GLM 4.5 Air...
https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF
This one is to fat for the 64gig KS-LE-B though.
https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF
Curious on how/what "slow" is - I guess you are running these as pure CPU LLMs?
Could you maybe share a some stats on a simple question+answer and how long it took + token stats?
Just wondering how usable this is (even if one is prepared to wait)...
To chat? fast enough.
For agentic coding? Gotta take a long piss.
Anywhere from 2t/s to 20t/s, depending on model, density, size etc
How long does "why is the sky blue" take? On any model (only CPU)... just curious.
Depends on model.
New Qwen 35b takes 55s.
Old Qwen 80b takes 80s
LFM2 24b took 22s.
PM sent, let's do a baguette swap
Look, I am a cheapskate.
I don't want any EU KS-MYSTERY.
Its more expensive than a CA MYSTERY.
64gig cost me 10€ paid in USD right now.
The MYSTERY EU would be about 30€ for double memory, geh.
Thanks for the offer though.
i pay only 23eur...
...
but i love it
I have a server with LE-B BHS 64GB memory. I can trade it with you.
If you're a forum member looking to snag a bargain at a low price, please don't contact me.
How much
Your firstborn
what would you want to trade it for ?
Brother, make your own thread, this is my turf.
The boys were always "swapping baguettes" if you know what I mean...
DM you
Why not to use API? As it’s cheaper than host on dedicated server