Skip to content

Anthropic

Claude to gain advanced memory with Memory Files

Written by an AI model from the linked sources and checked against them by a second model. How we work

Image: testingcatalog.com

Anthropic is overhauling Claude’s long-term memory system. The current single fact file about each user will give way to a folder called Memory Files - a set of structured documents organized by topic and project.

The approach resembles a personal knowledge base. Claude will be able to reference specific files selectively, which should save on context window usage. Users will be able to browse and edit these notes at any time.

Separately, the company is testing a feature called Dreams to optimize memory in the background. The algorithm works like the rapid eye movement stage of sleep: it merges duplicates, removes outdated data, and resolves contradictions. Dreams is currently being tested in the API for Opus 4.7 and Sonnet 4.6 models.

Anthropic

Anthropic set to release Mythos 1 for security and code

Anthropic is prepping to release Mythos 1, a powerful vulnerability detection model previously kept in-house. It will integrate directly into Claude Code. This is a significant move. Developers will gain access to what was essentially a proprietary military-grade tool for protecting their code.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.