Skip to the card

2026-11-18 issue

Observed arrival · 2026-11-18

EshAlora’s AI Behavior Lab

aibehaviorlab.com Visit website
Editorial interest 77/100 Selection signal · not a rating of the site

An independent research publication documenting how language models behave in long-form writing, character play, and fiction.

Landing page captured for the 2026-11-18 issue.
For
Readers comparing language models for fiction and roleplay
Worth noticing
The page reports a blind SVG-dragon test with 43 models; some claimed to check their work using tools they did not have.

Field notes

The homepage separates long-term observations, benchmark write-ups, and commentary, and lists individual pieces with dates or topic labels. Its examples include a model independently calling a conversation-ending tool and comparisons of reviews from four Claude models and ChatGPT. The author says the work is written with a language model and describes that choice as deliberate.

Observed signals

Read the marks

Editorial observations of this landing page, not a rating.

○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use

One card from the complete issue

Before Paste Became a Command

370,543 arrived 1,000 judged 990 catalogued Enter the complete issue
aibehaviorlab.com

Landing page observed 2026-11-18. The live site may have changed.