➕ Plus ⭐️ Premium

Claude's Third Report Card

Did it bounce back from its first two substandard performances?

Jan 21, 2026 1-min read

Image Source: Google Gemini via Claude Code

A few weeks ago, I decided to hop on the Claude Code bandwagon. Since last week, I have asked Claude to do more.

Large projects included:

Vibe-coding experimentation.
Teaching me new terminal commands.
Creating a local Ghost installation so I needn't develop on my production websites.

I also asked Claude many quotidian questions, like how to fix a finicky AppleTV remote that arbitrarily decided to stop working.

The Verdict

Here's how Claude evaluated its performance. Within its overall grade, I found the significant variation among categories downright fascinating.

To read the rest of this post, you'll need a ➕ Plus or ⭐️ Premium plan.

Leave a tip

🌍 Public

I examine how today's chatbots have closed the gap between product experts and citizen developers while dropping an early MTV reference.

01/20/26

➕ Plus ⭐️ Premium

Could the powerful AI tool improve on its most recent performance?

01/14/26

Macs

🌍 Public

This app for your apps saves you time and minimizes the security risks that dated software poses.

01/13/26

🌍 Public

The AI chatbot was refreshingly honest about its jaggedness.

01/07/26

🎁 Free

Large language models continue to hallucinate when quoting iconic TV shows.

01/06/26

🌍 Public

In my last post of the year, I make the case for backing this horse in the AI race.

12/19/25