I came across an ~~sticker~~ article recently which talked about sites/tools that tell you what could run on your machine (canirun.ai and llmfit).
whichllm seems to tell what gives you the best results on your hardware.
this post was submitted on 25 Jun 2026
30 points (91.7% liked)
LocalLLaMA
4892 readers
11 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
Thanks!
I've had good results with this model on my 16GB card. gemma-4-26B-A4B-it-ultra-uncensored-heretic
I'd suggest using koboldcpp.
I run gemma4:26b in 16 GB of RAM. It's slow on my test rig with only 2 GB VRAM but it should fit 16 GB VRAM fine. I have one of those AMD BC-250 crypto mining units setup as a gaming rig, but my plan was to also run ollama on it. gemma4:26b was the model I planned to make the default. I haven't messed with it yet since I'm playing through my Steam catalog that was waiting for me to have a PC that could run them lol.
Qwen 3.6 35B. It's A3B so it fits with space to spare for context. Just make sure you have --cpu-moe.
Thanks! Why the --cpu-moe flag though? And can you share your parameter section (the after --server part), if you're running llama?
I use model preset file but that just contains the equivalent of command line options. Here's what I have for Qwen:
[Qwen3.6-35B-A3B-MTP-Q4_K_XL]
m = /models/Qwen3.6-35B-A3B-MTP/Qwen3.6-35B-A3B-UD-Q4_K_XL.gguf
mmproj = /models/Qwen3.6-35B-A3B-MTP/mmproj-BF16.gguf
spec-type = draft-mtp
spec-draft-n-max = 2
chat-template-kwargs = {"preserve_thinking": true}
temp = 1.0
top-p = 0.95
top-k = 20
min-p = 0.0
presence-penalty = 1.5
repeat-penalty = 1.0
I don't use cpu-moe because I have enough VRAM for the whole model. If you have 16GB VRAM, you add cpu-moe which makes llama.cpp put only the active layers on the GPU. Keeps the rest of them in system RAM. Then it swaps them around as needed. The result is lower but still decent speed. On my hw, this model does 90-100tps when fully in VRAM. When using cpu-moe, I think it falls to 40-50tps, if I remember correctly.