Urgent.News

What's breaking now, across thousands of outlets.

AI

The #1 row on this AI memory leaderboard is not a measurement

Bench'd (benchd.ai) calls itself the neutral benchmark authority for AI memory, and sells vendors a verification badge from $299 to $3,999.99 a month. I ran my memory system through their harness. Then I checked their board. Every number below is from their own site and repository, read on 2026-08-29, and every one takes seconds to verify. The numbers are fake Their three track leaders,…

The AI memory leaderboard's top-ranked entry is not a genuine measurement. The benchmark service, Bench d, sells a verification badge to vendors for $299 to $3,999.99 per month. After testing their system, it was discovered that the numbers displayed on their leaderboard are fake. The top three entries - Knowledge Brain, Agent Memory Letta, and Conversational Memory LangMem - were all marked as Community-Verified, but upon verification, it was found that there is no manifest anyone can download to prove this.

Additionally, the harness repository, which is the basis for the independent and reproducible claim, has not seen a commit since June 6, 2026, and there are five open issues filed by the same person. Furthermore, the trust page of Bench d claims they do not take payment from vendors, but their pricing page sells them leaderboard badges, creating a contradiction.

Overall, this AI memory leaderboard is not measuring what it claims to, and the numbers displayed are misleading.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

I built an AI incident responder that refuses to fix anything without asking

Built for the WeMakeDevs × TrueFoundry Agent Harness Hackathon. There are two kinds of "AI for incident response," and both of them are wrong. The first acts on its own.

  • Mayday AI incident responder combines speed of autonomous action with safety of human approval
  • Proposes single fix with root cause, impact, and ruled-out options before human approval
  • 401 HTTP error prevents bypassing approval process

Argentina most exposed to AI impact in Latin America, study finds

The risk is related to high urbanization and the prevalence of the service industry, with Buenos Aires leading the wa La entrada Argentina most exposed to AI impact in Latin America, study finds se…

  • Argentina ranks highest in Latin America for AI exposure at 0.292
  • Service sector's 88% employment drives Argentina's AI vulnerability
  • AI adoption increases demand for supervision and judgment skills

More from Saturday 29 August →