New agentic memory framework uses 118K tokens per query. LangMem burn… | HappeningNow.news

Technology · 25 views

New agentic memory framework uses 118K tokens per query. LangMem burns through 3.26M.

Long-horizon reasoning exposes a core weakness in AI agents: context windows fill up fast, and retrieval pipelines return noise instead of signal.

Source AI Summary Published June 26, 2026 Brief Under 1 min brief
Story intelligence
Coverage Single outlet Single-outlet story
Views 25 Community interest
Brief read Under 1 min brief 77 words

AI Summary

Researchers at the National University of Singapore have developed a new framework for artificial intelligence agents called MRAgent. This framework differs from traditional approaches by allowing agents to dynamically process information, rather than relying on static retrieval and reasoning. The MRAgent framework uses a large number of tokens, specifically 118,000, per query to facilitate this dynamic processing. In contrast, another framework, LangMem, reportedly consumes a significantly larger amount of tokens, approximately 3.26 million, to achieve similar results.

AI summaries can be wrong sometimes—always verify important details using the source article.

How AI & Automation are used
Read original at Venturebeat

Coverage Context

Part of 26M coverage 3 tracked stories
Explore 26M

More from Technology

Continue reading recent Technology coverage

Support HappeningNow

Independent AI-powered news analysis is reader-supported. Your contribution helps cover infrastructure, summaries, and continued platform development.

Support HappeningNow

Report an issue with this page