loved it!!
#1
by gopi87 - opened
./bin/llama-server
--model "/home/gopi/deepresearch-ui/model/Qwen3.5-122B-A10B-APEX-I-Balanced.gguf"
--n-cpu-moe 63
-ngl 99
--ctx-size 250000
--tensor-split 1,1
--threads 28
--threads-batch 28
-np 1
-ub 5500
-b 5500
--host 0.0.0.0
--jinja
--port 8081
tested it used 250k fully