Through my current effort to port [OimoPhysics](https://github.com/saharan/OimoPhysics) from Haxe to Go, I created a tool that uses an LLM to review the conversion module-by-module compiling reports about any issues it finds. I expected a lot of stupid typos that could lead to hours of wasted time debugging and the reviewer did not disappoint.. what surprised me more is that it's finding bugs in the original OimoPhysics code!

Here's a bug the reviewer caught that I found amusing+horrifying at the same time 😅

lol not even two days after I started depending on gemini 2.5 pro as the first time I depend on an LLM directly in my dev, Google basically kills the freetier for it (without announcement of course)..
https://www.reddit.com/r/GoogleGeminiAI/comments/1pg121u/is_free_gemini_25_pro_fried/

For now 2.5 Flash seems to still work, but that could just be a matter of time, I think I'll switch to another LLM or do my original plan of hosting a local one on my gaming PC.

@sirdimples that pattern is never going to change
@jds yeah, looking forward to affordable LPUs or whatever they'll be called