Skip to the card

2026-11-18 issue

Observed arrival · 2026-11-18

CompactBench puts AI memory loss through repeat compaction

compactbench.com Visit website
Editorial interest 83/100 Selection signal · not a rating of the site

An open benchmark tests how much information AI context-compaction methods retain after five rewrites under a fixed 1,024-token memory budget.

Landing page captured for the 2026-11-18 issue.
For
AI agent and memory-system developers
Worth noticing
Its example catches a method retaining Marlin’s old go-live date after compaction four, despite a later correction.

Field notes

The sample places a one-time date correction among code, logs, and figures from other projects, then shows a sectioned-summary method keeping the superseded date after compaction four. The listed failure families also test scoped rules, reference resolution, and whether work marked done is actually only in review. The page says leaderboard scores are family-balanced and include a 95% interval.

Observed signals

Read the marks

Editorial observations of this landing page, not a rating.

○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎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
compactbench.com

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