Artificial Analysis - Thorsten Meyer AI

Reality Check

FrontierCode: Would a Maintainer Merge Your AI’s Pull Request?

Glowing glass gateway with streams of code passing through and a luminous branch growing above it

FrontierCode grades AI patches like a maintainer would. Claude Opus 5.5 leads at 52.2% but fails 42% of attempts; GPT-6.1 Sol scores 49.1% for $0.81 per task against Opus’s $10.15.

Thorsten Meyer
October 13, 2026

View Post

Reality Check

Muse Wonder Full Joins the Frontier. Muse Wonder Charges Like It Did Too.

Glowing cobalt and turquoise figure rising above a landscape of luminous data charts

Meta closed the capability gap. Muse Wonder Full scores 56 and matches Claude Code on coding agents. But Muse Wonder costs $4.52 per index task, compared with $0.72 for GPT-6.1 Sol at the same score.

Thorsten Meyer
October 13, 2026

View Post

3D harbour promenade at sunrise with a calendar-week gate over the path, from the Altersteilzeit browser game

AI & Work

Five Days, Sixteen Versions, Nineteen Games: What I Learned Building a Retirement Countdown with an AI Coding Agent

A one-minute film became a daily countdown, nineteen 3D games, a narrator in twelve…

Thorsten Meyer
October 13, 2026

Reality Check

FrontierCode: Would a Maintainer Merge Your AI’s Pull Request?

FrontierCode grades AI patches like a maintainer would. Claude Opus 5.5 leads at 52.2%…

Thorsten Meyer
October 13, 2026

Row of glass pedestals with glowing orbs: a small cobalt orb in the foreground spotlight, a much larger turquoise orb on a pedestal behind it

Reality Check

I Tested 71 AI Model Configurations on My Own News Pipeline. The Best One Isn’t the One I Run.

GPT-6 Astra wrote the best articles in my own 71-configuration test. GPT-6 Luna writes…

Thorsten Meyer
October 13, 2026

Reality Check

Muse Wonder Full Joins the Frontier. Muse Wonder Charges Like It Did Too.

Meta closed the capability gap. Muse Wonder Full scores 56 and matches Claude Code on…

Thorsten Meyer
October 13, 2026