Letta / MemGPT vs Zep & Graphiti
Letta / MemGPT vs Zep & Graphiti: side-by-side comparison of two dedicated memory layers systems — architecture, taxonomy, license, pricing, MCP/A2A support, and direct edges.
Cost & capability
| Letta / MemGPT | Zep & Graphiti | |
|---|---|---|
| Capability band | competent | competent |
| Capability composite | 68 | 68 |
| Cost tier | free | free |
| $/Mtok input | 0 | 0 |
| $/Mtok output | 0 | 0 |
| Use cases | Long Running Session, Memory Augmented Chat, Multi Agent Coordination | Long Running Session, Memory Augmented Chat, Multi Agent Coordination, Analytical Summarization |
Where they differ (14)
Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.
| Letta / MemGPT | Zep & Graphiti | |
|---|---|---|
| Use cases | Long Running Session, Memory Augmented Chat, Multi Agent Coordination | Long Running Session, Memory Augmented Chat, Multi Agent Coordination, Analytical Summarization |
| Type | Tiered: core / recall / archival | Bi-temporal knowledge graph |
| Created | 2023-10 | 2024-08 |
| Latest release | Letta Code v0.34.30 (npm @letta-ai/letta-code, 2026-10-29); PyPI letta v0.34.12 (2026-10-12); last letta-ai/letta server tag v0.16.8 (May 2026) | v0.30.0 (2026-09-01) (graphiti-core; MCP server mcp-v1.1.0 same day) |
| GitHub | letta-ai/letta-code — ~3.7k stars, Apache-2.0 (current source: agent harness, terminal UI, App Server); letta-ai/letta (~25.3k stars, not archived) is now a README pointing to letta-code, with the retired Letta V1 API server on its archive branch | getzep/graphiti — ~32k stars, last commit 2026-09 |
| Pricing | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom | Free 10k credits/mo; Flex $125/mo ($104 annual) 50k credits; Flex Plus $375/mo ($312 annual) 200k credits; Enterprise custom |
| Funding | $10M total $70M val Seed · 2024-09 | $3M total Seed (additional) · 2024-04 |
| Backend storage | Postgres + pgvector | Postgres + Neo4j (Graphiti) |
| Deployment | Both | Both — Zep Cloud managed; Graphiti OSS self-hosted (Neo4j/FalkorDB) |
| API surface | REST, SDK: Python, TS | REST, SDK: Python, JS/TS, Go |
| Multi-tenancy | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation | Logical namespace per user/session; AWS VPC self-hosted option for full data residency; HIPAA BAA on Enterprise plan |
| MCP | native (first-party) — Letta serves & consumes MCP | native (first-party) — Graphiti MCP server |
| Optimised for | LLM-as-OS paradigm; agent autonomy in managing its own memory | memory operation tracing + drift / poisoning detection |
| Anti-fit | no anti-fit explicitly stated | not for use cases that don't run agent workloads in production |
At a glance
| Letta / MemGPT | Zep & Graphiti | |
|---|---|---|
| Section | Dedicated memory layers | Dedicated memory layers |
| Tier | T1 | T1 |
| Type | Tiered: core / recall / archival | Bi-temporal knowledge graph |
| Created | 2023-10 | 2024-08 |
| Latest release | Letta Code v0.34.30 (npm @letta-ai/letta-code, 2026-10-29); PyPI letta v0.34.12 (2026-10-12); last letta-ai/letta server tag v0.16.8 (May 2026) | v0.30.0 (2026-09-01) (graphiti-core; MCP server mcp-v1.1.0 same day) |
| License | Apache-2.0 | Apache-2.0 |
| GitHub | letta-ai/letta-code — ~3.7k stars, Apache-2.0 (current source: agent harness, terminal UI, App Server); letta-ai/letta (~25.3k stars, not archived) is now a README pointing to letta-code, with the retired Letta V1 API server on its archive branch | getzep/graphiti — ~32k stars, last commit 2026-09 |
| Pricing | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom | Free 10k credits/mo; Flex $125/mo ($104 annual) 50k credits; Flex Plus $375/mo ($312 annual) 200k credits; Enterprise custom |
| Funding | $10M total $70M val Seed · 2024-09 | $3M total Seed (additional) · 2024-04 |
| Backend storage | Postgres + pgvector | Postgres + Neo4j (Graphiti) |
| Deployment | Both | Both — Zep Cloud managed; Graphiti OSS self-hosted (Neo4j/FalkorDB) |
| API surface | REST, SDK: Python, TS | REST, SDK: Python, JS/TS, Go |
| Embedding | multiple supported | multiple supported |
| Multi-tenancy | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation | Logical namespace per user/session; AWS VPC self-hosted option for full data residency; HIPAA BAA on Enterprise plan |
| MCP | native (first-party) — Letta serves & consumes MCP | native (first-party) — Graphiti MCP server |
| A2A | not documented publicly | not documented publicly |
| OpenTelemetry | not documented publicly | not documented publicly |
| Optimised for | LLM-as-OS paradigm; agent autonomy in managing its own memory | memory operation tracing + drift / poisoning detection |
| Anti-fit | no anti-fit explicitly stated | not for use cases that don't run agent workloads in production |
Taxonomy
| Axis | Letta / MemGPT | Zep & Graphiti |
|---|---|---|
| storage | kv | graph |
| retrieval | similarity | graph-traversal |
| persistence | long-term | long-term |
| update | agent-controlled | append-only |
| unit | episode | episode |
| governance | inspectable | auditable |
| conflict | llm-arbitrate | bi-temporal |
Pros & cons
Letta / MemGPT
Pros: Most architecturally principled — explicit OS-style tiering makes capacity behavior predictable; UC Berkeley origin + active v1 maturity.
Cons: Tier configuration adds operational complexity vs flat vector memory; latency is sensitive to paging decisions on long histories.
Zep & Graphiti
Pros: Bi-temporal graph captures event time + ingestion time, making contradiction tracking and chronological reasoning correct by construction.
Cons: KG storage is heavier than vector for the same data volume; smaller funding base than Mem0 ($2.3M vs $24M).