@philkellr I'm considering buying another 3060 to double the VRAM to 24 GB and maybe another pack of "cheap" slow DDR4 RAM in order to Run Qwen 3.8 flash next.
Seeing in which direction local LLMs are going, I'm getting Opus 4.5 Opus vibes.
Honestly, Iโve been considering streaming my coding sessions on private projects after work.
Few things have helped me progress faster than making myself vulnerable to open feedback.
A tech savvy but non coder colleague just told me he pointed GPT to our user docs and let it create a python script to work with our API.
I love what we can do now โค๏ธ.
The moment you touch a relative small niche in programing, the vibe coding approach fails. How dare them making me learn how to normalize my inputs for byte-level BPE myself ๐ข.