Which LLM Is Best for eCommerce Tasks?

Imagine walking into your favourite burger joint and asking for sushi…

Okay, they might entertain you, and you might get something edible.

But you are in the wrong place for the job.

This is exactly what most of us are doing with AI right now.

This week, Google announced Gemini 3 Deep Think and, before that, OpenAI rolled out GPT-5.3-Codex, and Anthropic rolled out Claude’s Opus 4.6.

And it has reignited the same argument:

Which one is best?

Making us all feel like we have to pick a side and stick with it.

And I was guilty of this up until about 3 months ago.

So here’s my opinion…

You do not need to pick sides with an LLM.

Yes, the models are getting better and faster and more powerful.

But they are not all getting better in the same way.

Each team has a different vision of what “best” means.

• Some are trying to be the best thinking partner.

• Some are aiming for speed and cost.

• Some are obsessed with coding and building tools.

• Some want to live inside your documents, email, and calendar.

• Some are doubling down on safety and control.

• Some are aiming at the enterprise market

• Some are aiming at the mass market

And they are also learning from how everyone is using their tool, collecting data and adjusting accordingly.

Naturally, this is all making each of these companies adapt and react in different ways.

And the differences are starting to appear and become a little more obvious, too.

So if you stay loyal to one model for everything, you will hit points where it feels like:

• My prompts are rubbish.

• AI is overhyped.

• AI cannot do what people are saying it can do.

But AI isn’t the problem; you are just “asking for sushi in a burger joint”.

These AI companies are just like your business.

They all have different strengths and weaknesses.

And ultimately, you need to use the one that's best for the “job to be done”.

Chat soon,

Peter