{
  "schema_version": "1.1",
  "id": "s13:https://arxiv.org/abs/2609.05339v1",
  "slug": "2609-05339v1-06dws1c",
  "url": "https://feed7.dev/p/2609-05339v1-06dws1c",
  "title": "Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability",
  "why_included": "Agent memory can break across model upgrades even when the store is unchanged. Test migrations by direction, rebuild embeddings completely, and retain raw histories for repair.",
  "summary": "Across **48 synthetic histories**, fixed-schema knowledge graphs barely changed after a writer-model swap, while compressed notes moved by **+9.91 or −13.28 percentage points** depending on migration direction. A half-migrated RAG index gained 4.96 points versus **11.90 points with full re-embedding**.",
  "practical_implication": "Treat model upgrades as memory migrations: test each model direction separately, keep embedding spaces isolated, and re-embed the full index. Prefer normalized schemas where practical, and retain source histories instead of relying only on model-written notes.",
  "agent_context": "Across **48 synthetic histories**, fixed-schema knowledge graphs barely changed after a writer-model swap, while compressed notes moved by **+9.91 or −13.28 percentage points** depending on migration direction. A half-migrated RAG index gained 4.96 points versus **11.90 points with full re-embedding**.\n\nTreat model upgrades as memory migrations: test each model direction separately, keep embedding spaces isolated, and re-embed the full index. Prefer normalized schemas where practical, and retain source histories instead of relying only on model-written notes.\n\nThe evidence covers two open-weight models below 10B parameters and synthetic histories with exact answers. Store-only note repair missed the 90% recovery target in every case; raw history enabled recovery in **34 of 48 cases** for one tested direction, not universally.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2609.05339v1",
    "published_at": "2026-09-04T16:44:17.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "agent",
  "domains": [
    "coding",
    "data"
  ],
  "topics": [
    "agent-memory",
    "retrieval",
    "agent-reliability"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "The evidence covers two open-weight models below 10B parameters and synthetic histories with exact answers. Store-only note repair missed the 90% recovery target in every case; raw history enabled recovery in **34 of 48 cases** for one tested direction, not universally."
  ],
  "connected_context": {
    "meaning": "This makes model portability a missing reliability constraint across the prior memory designs. It favors fixed, normalized, inspectable representations over model-authored compression when upgrades are expected, and shows that retrieval continuity requires full embedding migration rather than a mixed index. The limited models and synthetic histories narrow this to a concrete migration warning, not a universal ranking of memory architectures.",
    "corpus_size": 703,
    "generated_at": "2026-09-08T10:04:45.713Z",
    "connections": [
      {
        "title": "thedotmack/claude-mem",
        "source_name": "GitHub",
        "source_url": "https://github.com/thedotmack/claude-mem",
        "feed7_url": "https://feed7.dev/p/claude-mem-1uba088",
        "reason": "Claude-Mem’s compressed persistent observations create the kind of model-written memory whose portability this study finds direction-dependent, strengthening the case for retaining source history alongside summaries."
      },
      {
        "title": "UniMem: Complementary Episodic-to-Parametric Memory for Boundary-Agnostic Task Streams",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2607.26017v1",
        "feed7_url": "https://feed7.dev/p/2607-26017v1-1opv1da",
        "reason": "UniMem’s learned consolidation raises an additional portability question: recurring knowledge moved into model parameters may be harder to preserve across model replacement than normalized episodic records."
      },
      {
        "title": "Intelligence + Continual Learning = Expertise — Yu Su, NeoCognition",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=I6aiEf3aEFQ",
        "feed7_url": "https://feed7.dev/p/intelligence-continual-learning-expertise-yu-su-neocognition-0f7dwfd",
        "reason": "The continual-learning vision depends on reusable state surviving upgrades; this study supplies a concrete stability requirement and shows that raw histories can sometimes recover knowledge that compressed notes alone cannot."
      },
      {
        "title": "MemLens: A Value-Aware Memory Management System with Interactive Analytics for LLM-based Agents",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2607.25992v1",
        "feed7_url": "https://feed7.dev/p/2607-25992v1-0q2kxfl",
        "reason": "MemLens treats retention as a quality and cost optimization, while this study adds migration robustness as another criterion for deciding which representations and source records to preserve."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-09-04T16:44:17.000Z",
  "modified_at": "2026-09-04T16:44:17.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2609-05339v1-06dws1c",
    "json": "https://feed7.dev/p/2609-05339v1-06dws1c.json",
    "markdown": "https://feed7.dev/p/2609-05339v1-06dws1c.md"
  }
}