Are there any free models as good as the commercial ones we can run on our own hardware?
LLMs up and running but runs very slowly with Nvidia 16GB GPU and 32 cores on my laptop.
K3 and GLM 5.3 flash are pretty mid by comparison when you consider how much more it will take to run them
On the cheap end you can do something like qwen 3.8 27B with 24GB on a laptop, and a very minimal linux configuration. That's more similar to my previous setup.
You can also do a lot more with mixed setups like cpu-moe. I could probably run GLM 5.2 on this hardware with the right setup, albeit without LoRA. I know people have been doing this to run qwen 3.8-flash-next with 32GB cards.