GPT-5.5 on the benchmarks still seem to perform better than this

Plus the vibe of the gemini models are so weird particularly when it comes to tool calling

At this point I kinda need them to shock me to make the switch