In this post, I’ll share my experience with Claude Code after my first month.

On the last day of May, I subscribed to Claude Max for Claude Code. I started with the $100 USD 5× plan, then after a few days I upgraded to the $200 20× plan. So far it's mostly a positive experience with minor daily frustrations.

Then on Jul 2, 2025, after having been a subscriber to the $20 Cursor Pro plan since March, I turned off usage-based pricing in Cursor, then unsubscribed completely (we’ll talk about that in a separate post).

Just FYI I'm currently subscribed to Claude Max 20× ($200), ChatGPT Plus ($20), and Google AI Pro ($20) at the same time. I had pretty good reasons to not keep Cursor for another month.

Update 1 (Jul 5): updated the Claude Code prompt usage figures

Update 2 (Jul 9): updated to include a section to more clearly showing usage limits

Who am I?

I have no formal training in coding and I’ve never been a developer or considered myself as one, not even after months of using AI-assisted coding tools at this point. I learned Python years ago via Datacamp and at my universities’ in-person classes but only for statistical analysis.

I’m pushing myself to code now because I have an idea for a complicated multi-agent system for scientific research automation. I have tried Cursor, Windsurf, ChatGPT Plus (o1, o3-mini, o3), Gemini Advanced / AI Pro (Gemini 2.5 Pro), and Claude Code (Opus and Sonnet 4), and I’m still subscribed to the latter three.

I guess I would fall into the “vibe coder” camp (god I hate this term…) and I wouldn't object to that, but I can assure you that I'm trying my best to learn, apply, fail, adapt all at the same time.

The largest codebase I’ve had was about 35K LOC in Python and I reached that scale with Claude Code. Before that it was about 15K with Cursor (mostly built with Claude 3.7 Sonnet).