How I used LLMs in 2025 and which ones I prefer for specific tasks.

It is a year since my first ActrixFT blog and a majority of my 40 blogs over the year have covered my experiences of using LLMs for a range of tasks. I have learnt a lot since that first blog; Using an LLM with Public CPMI-IOSCO Quantitative Disclosures for CCPs, and more significantly there has been a lot of innovation in the capability and feature sets of LLMs.

Today’s article is a short review of my use in 2025.

Which LLM?

One goal I had for the year was to try out and compare each of the major Frontier Model providers; which I selected as OpenAI ChatGPT, Anthropic Claude and Google Gemini.

I continue to use all three, but over time my usage has cemented to:

• Claude for coding (from within VSCode)
• Gemini for research, analysis and images
• ChatGPT for refining text and learning

While these are not hard boundaries and each of the three could be used effectively for any of the above tasks, my preference has become established by daily use and what I read and consume about model capabilities.

For instance, the current LM Arena Leaderboard has Gemini-3-Pro ranked first and I have found myself using Gemini more and more, specifically after Gemini-3-Pro was released. Claude features such as Artefacts, MCP and Skills have been great for establishing its usefulness beyond coding. ChatGPT is the one I have used the least, probably influenced by my perception that OpenAI is much more consumer focused than business enterprise.

However, ChatGPT is the only one that offered to show me a review of my use for 2025 (a nice consumer feature), let’s take a look at that.

ChatGPT Use

My highlights as presented by ChatGPT, starting with a poem.

So picks up my analysis of the US Bank Holding Company Report FRY9C and building Dashboards, a theme in a number of blogs (see Banks).

Next, the top 3 themes, which also highlight my use for both Actrix corporate tasks and help in refining blogs.

A few stats:

My chat style, nice, is this flattery to get me to use more….?

And finally my archetype profile, is an Engineer.

What’s yours?

Just ask ChatGPT.

That’s All

That’s all for today.

I hope you made effective use of LLMs in 2025.

In work as much as personal life.

And I hope my blogs have been helpful.

It remains for me to wish you a Happy New Year.