Pi Code with Gemma 4 QAT

The agentic coding I have done so far has been with OpenCode, but I have been hearing about Pi Code so today I tried it out. It seems somewhat lighter than OpenCode. I thought maybe the Gemma 4 12B QAT model I have (with MTP too) might work better in Pi Code.

Pi Code

I installed the software on my Ubuntu Linux PC and figured out how to use local models with it—one must add a models.json file in the hidden "pi" directory in one's home folder (everything is always as complicated as possible I guess).

Working with Gemma 4 12B, I found:

• Pi Code seems to be lighter than OpenCode, and this helps for people using local models on mid-tier GPUs.

• Gemma 4 seemed to work better in Pi Code, but I found that typing "new" to start a new session was helpful as it reduced model confusion.

• Tool calling worked, but (as in OpenCode) the model had trouble calling tools (like when editing files).

I made some worthwhile changes to my Rust program with the setup, but as with most local AI stuff further improvements are needed. I am not sure whether I will prefer Pi or OpenCode; they seem similar. Pi Code will likely be faster, and for local agentic programming, speed is important.