Mem0

C+▲ 0 vs last quarter

Persistent memory layer for AI agents; v2.0 algorithm (April 2026) adds single-pass extraction, multi-signal retrieval, entity linking, and temporal reasoning.

Rate this tool Add to compare

Dimension breakdown

Score · confidence

UX / DX

50% conf75

Capability

50% conf90

ReliabilityIncomplete data at this time

Value

50% conf70

CommunityIncomplete data at this time

Enterprise / Compliance

50% conf45

Autonomy

50% conf35

Integration

50% conf45

Sources blend review platforms, community sentiment, the Signal Radar and practitioner ratings. Scores re-blend each quarter.

Persistent memory layer for AI agents; v2.0 algorithm (April 2026) adds single-pass extraction, multi-signal retrieval, entity linking, and temporal reasoning. Vendor claims 92.5 LoCoMo / 94.4 LongMemEval but benchmarks remain independently unreproduced — practitioners report reproduction failures and one independent analysis found long-context GPT-5-mini outperforming Mem0 by 35.2/33.4 pts.

Strong ecosystem: OpenMemory MCP (Claude Code/Cursor/Windsurf), 21 framework integrations, AWS Strands exclusive memory provider, Trend Micro reference, ~57K GitHub stars, 80K+ developers, team grown to ~22. SOC 2 Type II audit underway (Type I certified, HIPAA-ready, BYOK, self-hosted Docker, Trust Center live).

Caution

UNPATCHED HIGH-severity SQL/Cypher injection (#4875, GHSA-5gv3-2fv6-jvhx, CVSS 8.1 Neptune / 6.5 PGVector+MySQL, OPEN, fix PR #4878 not merged as of 2026-06-19, public PoCs, no CVE, no active exploitation — compliance dim reduced 12->9; critical-security-vuln cap NOT applied; reassess when patched); reliability multi-signal persists (#4573 97.8% junk audit open, Scira AI operator migration to supermemory citing latency/scaling, Neo4j graph latency, OpenClaw plugin #4037).

Competitive pressure from Letta, Zep, Cortex, Supermemory, Hindsight.

Recommended

Use cases

Not yet assessed — this section fills in as ACES research covers the tool.

Score caps

Risk flags

Unvalidated benchmarks
trustConditional
Unvalidated benchmark claims
Caps Autonomy at 70
Removed when — Independent benchmark validation (SWE-bench, Aider leaderboard, etc.) published
Reliability complaints
trustTemporary
Widespread reliability complaints (breaks often, unreliable output)
Caps Autonomy at 60
Removed when — 90+ days of improved reliability with community acknowledgment

Assessment

Status rationale

Assessed — HOLD. Substantial multi-source evidence (architecture, integrations, named enterprise references, compliance posture) supports Assessed, but two active caps and an open HIGH-severity security vuln prevent Validated: (1) unvalidated-benchmarks — no independent third-party replication of v2.0 LoCoMo/LongMemEval, with active reproduction failures and an independent refutation (long-context GPT-5-mini outperforms Mem0); (2) reliability-complaints — multi-source (GitHub #4573 junk audit open, Scira AI operator migration to supermemory, OpenClaw plugin #4037, Neo4j graph latency); (3) NEW unpatched HIGH SQL/Cypher injection #4875 (CVSS 8.1 Neptune) is an open blocker for Validated even though it is below the Critical threshold for the critical-security-vuln dimension cap. Revisit when #4875 is patched and v2.0 benchmarks see independent validation.

Watch for

Movement triggers

#4875/GHSA-5gv3-2fv6-jvhx patched (PR #4878 merged, patched release ships) AND no new critical/high CVE — compliance can be reassessed upward from 9 once patched; independent third-party LoCoMo/LongMemEval validation of v2.0 figures, SOC 2 Type II certification completed, GitHub #4573 closed with verified production junk-rate metrics, sustained reliability with no new operator-migration signals, SSO/SCIM + training-opt-out documented, Series B funding.

Down

#4875 escalated to Critical (CVSS 9.0+) or exploited in the wild, new direct-codebase critical CVE, v2.0 benchmark claims formally refuted by independent replication, reliability complaints escalate beyond current multi-source baseline, competitive displacement accelerates (e.g., AWS Strands SDK exclusivity flips), team attrition or Series A runway concerns.

Caution