Glastonbury 2025 Main Lineup Cheat Sheets – 🏆Awards and 🥇Results

Overview

Time to see which GenAI helped us music lovers most to unearth the Glasto lineup gems.

I made tons of them (see my last post) and checked for which GenAI made the best cheat sheets

After crunching and random sampling, I can rate the results and award some winners

†About the Sampling and Methodology

I chose some artists at random that I did not know and mixed in artists I did know. I did a deep dive where the results differed between the outputs. I looked at the artists’ Wikipedia. The artists I knew are in bold below:

Parcels
Rod Stewart
Joy Crookes
The Maccabees
The Prodigy
THE BRIAN JONESTOWN MASSACRE

🥇Results🥇

Here are the results as a table: (click to enlarge)

Glastonbury 2025 Artist Info Cheat Sheet - Awards Table
Glastonbury 2025 Artist Info Cheat Sheet – Awards Table

🏆🏆Awards🏆🏆

Overall Winner

🏆Claude Sonnet 3.7 without Search🏆

Claude had the best Blurb. Check it out for youself at the dedicated page for the Claude results. The words flowed, and “terse” was not interpreted as “abrupt”. Claude was also the only one that added useful extra interesting tidbits. For example The Brian Jonestown Massacre had a description about their rivalry with The Dandy Warhols and the documentary made about that rivalry. This is magic dust.

Here’s what Claude had to say about the acts:

The Prodigy: “Pioneering electronic dance act known for aggressive sounds and energetic live performances that bridged rave culture and rock attitude.”

Parcels: “Australian five-piece based in Berlin creating sophisticated disco-funk with impeccable harmonies and retro-futuristic production.”

Rod Stewart: “Iconic raspy-voiced singer with over five decades in music, known for seamlessly blending rock, folk, and pop styles.”

The Brian Jonestown Massacre: “Cult psychedelic band led by Anton Newcombe known for prolific output, 60s influences, and the documentary “Dig!” documenting their rivalry with The Dandy Warhols.”

Claude also placed well for Notable Works Accuracy and okay Result Speed

Most Accurate

🏆Grok with DeepSearch🏆

🥈Google Gemini🥈

Grok was the only one to get the dates correct for the randomly selected artist Joy Crookes. I checked on Joy’s Wikipedia page and it clearly shows 2013 is the beginning of Joy’s career at an impressive 14yo. This tidbit is the type of thing we want from GenAI.

Grok also nailed Prodigy’s Notable Works, including “Fat of the Land” which has very many seminal tracks. Gemini also did well.

‡Joy Crookes’ Wikipedia Page showing 2013

Fastest

Google Gemini

Grok with DeepSearch

Both these were very fast under 30 seconds.

Awards Summary

So congratulations to Claude, Grok and Gemini.

The Failures

Several LLMs failed the task.

Chat GPT-4o only got part way through on all days. My hunch is token window size is exhausted by the list of artists. This is despite the fact I am a $20pcm Plus subscriber. o3-mini and o3-mini-high had no problems

ChatGPT 4.5 with Search also got part way through as well. Clearly because I am on the Plus subscription 4.5 is only given to me in small sample amounts. This is made clear by ChatGPT that I am getting a sample. Again I put the non completion down to the input list of Artists overflowing the token window.

Google Gemini experienced a visible problem in one attempt. I tried again and it worked. The failed attempt generated results for all the artists (good) and the results were markdown tables (good) however the result was missing important rendering information and so it rendered as fixed text table. Not terrible. All LLMs have trouble following instructions, even those given by the platform itself. So this is to be expected every so often. That’s why I felt comfortable giving Google Gemini a second attempt.

Google Gemini first attempt rendered as text rather than markdown
Google Gemini – The first attempt rendered as text rather than markdown

Coming Soon…

Coming soon are more generative AI comparisons

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *