# Supersede: Diagnosing and Training the Memory-Update Gap in LLM Agents arXiv: https://arxiv.org/abs/2606.27472 ยท HTML: https://mlsanity.com/paper/2606.27472/ ## Key facts - Cited by 20 papers in the map (data through 14 Oct 2026). - Placed in Agent memory systems, one of 16 directions in LLM agents and tool use. - Builds on 14 of its 22 references that are in the map. - First posted on arXiv 25 Jun 2026. ## Placed in - LLM agents and tool use / [Agent memory systems](https://mlsanity.com/agents/agent-memory-systems/) ## Builds on (most cited first, 14 of 14) - [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024) - [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024) - [Evaluating Very Long-Term Conversational Memory of LLM Agents](https://mlsanity.com/paper/2402.17753/) (2024) - [Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory](https://mlsanity.com/paper/2504.19413/) (2025) - [LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory](https://mlsanity.com/paper/2410.10813/) (2024) - [Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions](https://mlsanity.com/paper/2507.05257/) (2025) - [MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent](https://mlsanity.com/paper/2507.02259/) (2025) - [MemBench: Towards More Comprehensive Evaluation on the Memory of LLM-based Agents](https://mlsanity.com/paper/2506.21605/) (2025) - [MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks](https://mlsanity.com/paper/2602.16313/) (2026) - [Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents](https://mlsanity.com/paper/2601.01885/) (2026) - [From Recall to Forgetting: Benchmarking Long-Term Memory for Personalized Agents](https://mlsanity.com/paper/2604.20006/) (2026) - [LongRLVR: Long-Context Reinforcement Learning Requires Verifiable Context Rewards](https://mlsanity.com/paper/2603.02146/) (2026) - [Memory-T1: Reinforcement Learning for Temporal Reasoning in Multi-session Agents](https://mlsanity.com/paper/2512.20092/) (2025) - [Question Answering under Temporal Conflict: Evaluating and Organizing Evolving Knowledge with LLMs](https://mlsanity.com/paper/2506.07270/) (2025) ## Built on by (most cited first, 20 of 20) - [When Memory Updates but Behavior Does Not: Repairing Implicit Stale Dependencies in Personalized Agent Responses](https://mlsanity.com/paper/2608.01619/) (2026) - [Ground Truth First: A Longitudinal Evaluation Instrument for Agent Memory, and the Tenure Crossover in Memory-Architecture Rankings](https://mlsanity.com/paper/2607.21962/) (2026) - [When Stale Constraints Go Unchecked: Budgeted Verification Failures in Inherited Agent Memory](https://mlsanity.com/paper/2608.25553/) (2026) - [Agents in the Large: Perception-Centered Architecture for Persistent Agents](https://mlsanity.com/paper/2608.30478/) (2026) - [The Memory Trust Gap: Capability-Dependent Failures in Persistent-Memory Agents](https://mlsanity.com/paper/2609.01852/) (2026) - [ERRAND: Budgeted Maintenance of Agent Memory](https://mlsanity.com/paper/2609.29545/) (2026) - [Retrieved but Not Delivered: Multimodal Memory Delivery for Long-Term Agents](https://mlsanity.com/paper/2609.32590/) (2026) - [Transfiver: Human-AI Co-Inference through a Shared Editable State](https://mlsanity.com/paper/2609.03797/) (2026) - [ROAM: Robust Organization of Atomic Memories for Agents through Semantic Relations](https://mlsanity.com/paper/2609.09778/) (2026) - [Self-Cleaning and Captured Anyway: One Measured Primitive for Error in a Store an Agent Writes to Itself, and What a Falling Score Actually Measures](https://mlsanity.com/paper/2609.25052/) (2026) - [When Users Change Their Minds: Measuring and Repairing Intent Drift in LLM Agents](https://mlsanity.com/paper/2609.32520/) (2026) - [RoutePrism: Tracing Construction Order Effects in Agent Memory](https://mlsanity.com/paper/2609.34160/) (2026) - [Authority Before Utility: Non-Compensatory Control for Persistent LLM Memory](https://mlsanity.com/paper/2609.37474/) (2026) - [When Context Changes: Understanding Update Failures in LLMs](https://mlsanity.com/paper/2609.38866/) (2026) - [The Delegation Danger Band: Why Mid-Capability Sub-Agents Over-Trust Inherited Stale State](https://mlsanity.com/paper/2610.00041/) (2026) - [When Evidence Changes: Evaluating Memory Repair and Re-reading in Language-Model Agents](https://mlsanity.com/paper/2610.03902/) (2026) - [Knowing the Store: What a Memory Backend Must Write Down Before an Agent Can Read It](https://mlsanity.com/paper/2610.04794/) (2026) - [StateWise: Diagnosing and Repairing Persistent Operational State Before Agent Actions](https://mlsanity.com/paper/2610.05241/) (2026) - [When Agent Context Goes Stale: Incoherence in Volatile Agent Context](https://mlsanity.com/paper/2610.05281/) (2026) - [AgentMemGate: Addressing Speculation Contamination in Conversational Assistant Memory](https://mlsanity.com/paper/2610.07707/) (2026) ## Abstract (arXiv) Large language model (LLM) agents operate over long, multi-session interactions in which facts change: a user moves, a price updates, a plan is revised. Acting correctly requires using the current value of a fact and discarding values that have been superseded. We isolate this ability on real conversational data and show that it is a distinct, unsolved failure. On the knowledge-update subset of LongMemEval, replacing an agent's full context with a bounded, self-maintained memory drops accuracy from 92% to 77% even on a frontier model (gpt-5.4), a gap that is statistically significant (paired McNemar p<0.005) and persists across model scale while full-context accuracy saturates near 92%. The bottleneck is therefore memory maintenance, not comprehension, and is not closed by a stronger model. We then ask whether this is merely an undersized memory, and find it is not: as the conversation grows 24x, accuracy falls further (from 68% to 28%), and granting the agent proportionally more memory yields no detectable recovery (28% to 28%, n=25). The failure scales with the length of the conversation, not the compression ratio. We release Supersede, an open reinforcement-learning environment (on the verifiers / prime-rl stack) that turns this measurement into a training signal: agents are rewarded for answering from the current value and penalized for stale ones. Finally, we close the loop and show the gap is trainable: GRPO fine-tuning a small open model (Qwen2.5-3B) on this environment nearly doubles its held-out supersession accuracy on real, unseen conversations (9.0% to 16.7%, a single run), along a monotonic checkpoint curve indicating the learned policy, not the harness, carries the gain. To our knowledge this is the first trainable environment whose reward targets temporal fact-currency, and the first evidence the supersession gap can be trained down, not only measured.