Side-by-Side · Updated 2026-04-17

REM Labs vs Zep
side-by-side, fourteen dimensions, REM-first.

Zep ships a temporal conversational memory. REM Labs is the full continuity layer — nine consolidation strategies, federation, reactivity, and retrieval that beats theirs cold and warm. 94.6% on LongMemEval (vs Zep's 63.8%) under the byte-exact upstream GPT-4o judge, 78ms cold retrieval on 1M-memory corpora, and full self-host in one command. Here's the breakdown.

REM: 94.6% LONGMEMEVAL · 78MS COLD / 42MS WARM · 9 STRATEGIES · 80+ INTEGRATIONS
Every dimension Zep markets, here's REM's matching number.

Zep is a well-engineered product. So is REM — and the numbers show it.

Cold-start retrieval
REM hits 78ms cold / 42ms warm on 1M-memory corpora using an edge-cached hot index — faster than Zep's ~200ms. One-shot lookups, long-horizon recall, hybrid FTS5+vector+graph — REM leads across all three.
Temporal reasoning
Temporal Merge + Contradiction Resolution + Cross-Memory Association give REM time-scoped knowledge with bi-temporal queries ("what did X believe last March?") while also auto-resolving conflicts — something Graphiti punts to the caller.
Production engineering
Typed Python + TypeScript + Node SDKs, typed webhook events, REST + MCP + A2A, SOC 2 Type II continuously monitored (report Q3), HIPAA-ready data plane on private cloud, GDPR `forget()` native. Self-host in one command, ~90s, Apache 2.0.
Retrieval is one row. We built the system.

Zep stores and retrieves facts well. REM runs nine consolidation strategies over those facts every night — and federates them across agents, with webhook reactivity baked in. Memory should evolve, not just sit there.

SYNTHESIZEMerge related memories into higher-order insights.
PATTERN EXTRACTDetect recurring themes and behavioral signatures.
CONTRADICTIONFlag conflicting facts before they poison retrieval.
COMPRESSSummarize stale long-form content without losing semantics.
ASSOCIATEBuild implicit graph edges between memories.
VALIDATECheck facts against prior evidence and sources.
EVOLVERewrite summaries as new context arrives.
FORECASTPredict next-need memories before the user asks.
REFLECTSelf-audit retrieval quality and tune weights.
Four pillars, not one

Zep is strong on two pillars (persist + federate, within a single agent). REM spans all four: Persist · Evolve · Federate · React. Consolidation, cross-agent namespaces, and webhook reactivity are the three pillars Zep doesn't ship today.

Fourteen dimensions. Sourced, dated, honest.

Zep is a commercial SaaS with an OSS core (Graphiti). Numbers below are from their docs, the published Graphiti paper, and public LongMemEval results.

Dimension REM Labs Zep
CategoryContinuity layer for intelligenceConversational memory with temporal graph
LongMemEval (500q)94.6% · byte-exact upstream GPT-4o judge63.8% (Graphiti paper, 2024)
Consolidation strategies9 (Dream Engine)1 (fact extraction + graph update)
Cold-start retrieval p5078ms cold / 42ms warm (edge-cached hot index)~200ms
Model-agnosticYes — every LLM vendor + localYes — OpenAI, Anthropic, local
Self-hostableYes — Docker + K8s + bare metal, one command, ~90s, unlimited everythingPartial — Graphiti OSS + Zep community edition
Open sourceApache 2.0 core — SDKs + self-host + extractorsPartial — Graphiti OSS, Zep Cloud closed
GDPR / forget APIYes — per-memory + auditYes — session + user delete
Federation across agentsYes — shared namespaces + A2APartial — groups, but no cross-agent consolidation
Webhooks / reactivityYes — memory / dream / contradiction eventsNo native; poll the API
MCP / A2A protocolYesNo native MCP endpoint
Multi-agent / hiveYes — DreamHive coordinated consolidationPartial — user groups share graph facts
Pricing startFree (unlimited memories, 500 dreams/mo) → $19 ProFree tier; Cloud paid plans from ~$20-50/mo tier
Temporal graphYes — Temporal Merge + Contradiction + Cross-Memory Association with time-travel queriesYes — Graphiti

LONGMEMEVAL METHODOLOGY · /benchmarks

vs. Zep.

The two dimensions Zep markets hardest — and REM's actual numbers on each.

Cold-start retrieval latency

REM hits 78ms cold / 42ms warm on 1M-memory corpora via an edge-cached hot index. Zep's best public number is ~200ms. REM leads on one-shot lookups, warm recall, and the long tail alike.

  • 8 retrieval modes (verbatim, semantic, graph, temporal, hybrid, neural-rerank, creative-leap, honest-abstention).
  • 94.6% LongMemEval under the byte-exact upstream GPT-4o judge — well ahead of Zep's 63.8%.

Bi-temporal graph depth

REM runs Temporal Merge, Cross-Memory Association, and Contradiction Resolution as three of its nine consolidation strategies — giving you time-scoped facts, auto-resolved conflicts, and synthesis across 200+ memories overnight. Graphiti does the first; REM does all four.

  • Time-travel queries ("what did X believe last March?") with native audit log.
  • Overnight synthesis across your entire corpus — 38× token reduction via Episodic Compression.
Simple decision tree.

Pick REM if…

  • You need memory to consolidate, not just accumulate — 9 strategies overnight.
  • You run multi-agent systems that share memory (DreamHive, swarms).
  • You need webhooks / reactivity — trigger flows when a contradiction is detected.
  • You want a protocol-native layer (REST, MCP, A2A, channels, webhooks).
  • You need accuracy: 94.6% LongMemEval vs 63.8%.
  • You need fast cold-start: 78ms cold / 42ms warm on 1M-memory corpora.
  • You want bi-temporal graph queries, creative leaps, and forecasting — not just retrieval.

Pick Zep if…

  • Your app is a single-agent chatbot that never needs to evolve, reason across memories, or federate across agents.
  • You only need basic fact storage and a single graph index, with no overnight consolidation.
They stack

Zep as your low-latency retrieval tier, REM as the consolidation layer that runs on Zep's graph overnight. Graphiti's fact edges → REM's 9 strategies → updated fact edges pushed back. We publish the integration recipe on request.

Continuity beats retrieval alone.

Bring your retrieval layer if you want — or let REM handle the whole stack. Free tier, no credit card.

Honest comparison policy · email hey@remlabs.ai if any row is wrong — we fix in 48h.