Gemini 2.5 Pro

My strange observation is that Gemini 2.5 Pro is maybe the best model overall for many use cases, but starting from the first chat. In other words, if it has all the context it needs and produces one output, it's excellent. The longer a chat goes, it gets worse very quickly. Which is strange because it has a much longer context window than other models. I have found a good way to use it is to drop the entire huge context of a while project (200k-ish tokens) into the chat window and ask one well formed question, then kill the chat.

That particular future exists now, it's just not evenly distributed. Gemini 2.5 Pro Thinking is already as good at programming as I am. Architecture, probably not, but give it time. It's far better at math than I am, and at least as good at writing.

On the flip side, twice I put about 800K tokens of code into Gemini and asked it to find why my code was misbehaving, and it found it. The logic related to the bug wasn't all contained in one file, but across several files. This was Gemini 2.5 Pro. A whole generation old.

I think this will be useful but not an important feature. I also prefer using svg over canvas. Since svg doesn't work well with animations without JS, I think it's not ideal for that. Last year, I needed example circuits for a quiz app, so I chose CircuitJS because it can be embedded in a webview and supports importing and exporting circuits as text. Initially, AI (Gemini 1.5 Pro 001) had trouble generating the circuit text, but after I provided some example descriptions and scripts I created on Falstad.com, it started getting it right. However, It couldn't generate complex circuits. I never got around to testing its output this year. Maybe AI has improved a lot since then. I had considered training(fine tuning) an LLM for this task, but I don't think that's the best approach. perhaps latest AI can now generate circuits more reliably with a simpler system prompt. edit: latest AI can also generate CircuitTikZ (LaTeX). I tested image to circuittikz on Google ai studio, and Gemini 2.5 pro Experimental 03-25 generated it in just two attempts. The results were pretty impressive.