This blog post discusses the newly released Google's Gemini 2.5 Pro experimental. We discuss the benchmarks, reactions, and capabilites.

Google has finally arrived

Some observations on the model

  • Gemini 2.5 pro is absolutely a beast in coding, perhaps the best model right now
  • They spent all the computing resources on training it on coding data and forgot to give it a distinct personality
  • Doesn’t do well on reasoning as well as Grok 3 (think) and Claude 3.7 Sonnet (thinking)
  • On par with 03-mini-high in general mathematics

If you’re a coder, you’ll absolutely love it, or else you will be fine with other frontier reasoning models (Deepseek r1, if you ask me)