Which model do you use and what are your specs? I ran a couple using an RTX5060 with 16gb and it’s too slow to be usable for larger models while the smaller ones are mostly useless.
I also have a 5060 (ti) with 16GB of RAM. I tend to use GPT-OSS:20B or Qwen3:14B with a context of ~30k. I have custom system prompt for my style of reponse I like on open web ui. That takes up about 14GB of my 16GB VRAM
But yeah it is slower and not as “smart” as the cloud based models, but I think the inconvenience of the speed and having to fact check/test code is worth the privacy and environmental trade offs
Ive had good success on similar hardware (5070 + more ram) with GLM-4.7-Flash, using llama.cpp’s --cpu-moe flag - I can get up to 150k context with it at 20ish tok/sec. I’ve found it to be a lot better for agentic use than GPT-OSS as well, it seems to do a much more in depth reasoning effort, so while it spends more tokens it seems worth it for the end result.
Which model do you use and what are your specs? I ran a couple using an RTX5060 with 16gb and it’s too slow to be usable for larger models while the smaller ones are mostly useless.
I also have a 5060 (ti) with 16GB of RAM. I tend to use GPT-OSS:20B or Qwen3:14B with a context of ~30k. I have custom system prompt for my style of reponse I like on open web ui. That takes up about 14GB of my 16GB VRAM
But yeah it is slower and not as “smart” as the cloud based models, but I think the inconvenience of the speed and having to fact check/test code is worth the privacy and environmental trade offs
Ive had good success on similar hardware (5070 + more ram) with GLM-4.7-Flash, using llama.cpp’s
--cpu-moeflag - I can get up to 150k context with it at 20ish tok/sec. I’ve found it to be a lot better for agentic use than GPT-OSS as well, it seems to do a much more in depth reasoning effort, so while it spends more tokens it seems worth it for the end result.