alt
Hacker News
julianlam
•
yesterday at 6:10 PM
•
0 replies
•
view on HN
Try llama.cpp and Qwen3.6-35B-A3B
Good balance of intelligence and speed.