1
2 Comments

Opus 4.5 + Jupyter Notebooks: A game changer for Python data engineers

I tested three AI models on the same data analysis task in VS Code with GitHub Copilot. Same prompt, same dataset, same questions.

Gemini was fastest but hallucinated conclusions. It claimed rush hours were dangerous when the data showed late night was worse.

Codex was accurate but slow and surface level.

Opus struggled with VS Code notebook tools at first, but produced the best output by far. Thorough cleaning, real insights, conclusions I could verify.

For Python engineers doing ad hoc analysis, this workflow is worth exploring.

What setups are you using for AI assisted data analysis? Curious what's worked and what hasn't. I work with a lot of data at work, so I'm quite interested in your perspective.

Full comparison with methodology on my blog.
https://blog.jackcreates.com/gpt-5-2-codex-vs-claude-opus-4-5-vs-gemini-pro-which-model-actually-codes-best/

What is your preferred model for data analysis?
  1. Opus 4.5
  2. ChatGPT-5.2-Codex
  3. Gemini Pro
Vote
posted to Icon for group Developers
Developers
on January 18, 2026
  1. 2

    Opus 4.5 looks like a real step forward — tighter Jupyter support always feels like finally unlocking smoother workflows instead of wrestling with clunky hacks.How it handles larger pipelines and versioning without slowing things down.

Trending on Indie Hackers
30 days ago I posted here with $0 revenue. Here's what actually happened next. User Avatar 99 comments I used $30,983 of AI tokens last month in Claude code on $200/mo plan User Avatar 88 comments my reddit post got 600K+ views. here's exactly what i did User Avatar 44 comments How to spot high-intent customers in 5 minutes, for free. User Avatar 40 comments I turned someone’s tweet into an app idea and it has made ~$3000 so far in 4 months. User Avatar 37 comments I Built a Habit Tracker SaaS Alone in 6 Weeks (No CS Degree, No Team). Here's Exactly How User Avatar 28 comments