Cognee vs Letta / MemGPT
Cognee vs Letta / MemGPT: side-by-side comparison of two dedicated memory layers systems — architecture, taxonomy, license, pricing, MCP/A2A support, and direct edges.
Cost & capability
| Cognee | Letta / MemGPT | |
|---|---|---|
| Capability band | competent | competent |
| Capability composite | 65 | 68 |
| Cost tier | free | free |
| $/Mtok input | 0 | 0 |
| $/Mtok output | 0 | 0 |
| Use cases | Long Running Session, Memory Augmented Chat, Analytical Summarization, Multi Agent Coordination | Long Running Session, Memory Augmented Chat, Multi Agent Coordination |
Where they differ (14)
Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.
| Cognee | Letta / MemGPT | |
|---|---|---|
| Capability composite | 65 | 68 |
| Use cases | Long Running Session, Memory Augmented Chat, Analytical Summarization, Multi Agent Coordination | Long Running Session, Memory Augmented Chat, Multi Agent Coordination |
| Type | Knowledge graph + ECL pipeline | Tiered: core / recall / archival |
| Created | 2023-08 | 2023-10 |
| Latest release | v1.6.5 (2026-10-16) | 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) |
| GitHub | topoteretes/cognee — ~31.2k stars, last commit 2026-10 | 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 |
| Pricing | Free $0 (1 workspace, 1M tokens); Standard $1 per 1M tokens + $5/workspace/mo; Enterprise BYOC custom | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom |
| Funding | $10M total Seed · 2026-02 | $10M total $70M val Seed · 2024-09 |
| Backend storage | hybrid (vector + graph) | Postgres + pgvector |
| Deployment | Both (OSS self-hosted; Cognee Cloud managed; Enterprise BYOC) | Both |
| API surface | REST, SDK: Python | REST, SDK: Python, TS |
| Multi-tenancy | Logical namespace per project/dataset; self-hosted OSS deployment | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation |
| MCP | native (first-party) — cognee-mcp | native (first-party) — Letta serves & consumes MCP |
| Optimised for | typed knowledge graph extraction (ECL pipeline) over RAG | LLM-as-OS paradigm; agent autonomy in managing its own memory |
At a glance
| Cognee | Letta / MemGPT | |
|---|---|---|
| Section | Dedicated memory layers | Dedicated memory layers |
| Tier | T1 | T1 |
| Type | Knowledge graph + ECL pipeline | Tiered: core / recall / archival |
| Created | 2023-08 | 2023-10 |
| Latest release | v1.6.5 (2026-10-16) | 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) |
| License | Apache-2.0 | Apache-2.0 |
| GitHub | topoteretes/cognee — ~31.2k stars, last commit 2026-10 | 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 |
| Pricing | Free $0 (1 workspace, 1M tokens); Standard $1 per 1M tokens + $5/workspace/mo; Enterprise BYOC custom | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom |
| Funding | $10M total Seed · 2026-02 | $10M total $70M val Seed · 2024-09 |
| Backend storage | hybrid (vector + graph) | Postgres + pgvector |
| Deployment | Both (OSS self-hosted; Cognee Cloud managed; Enterprise BYOC) | Both |
| API surface | REST, SDK: Python | REST, SDK: Python, TS |
| Embedding | multiple supported | multiple supported |
| Multi-tenancy | Logical namespace per project/dataset; self-hosted OSS deployment | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation |
| MCP | native (first-party) — cognee-mcp | native (first-party) — Letta serves & consumes MCP |
| A2A | not documented publicly | not documented publicly |
| OpenTelemetry | not documented publicly | not documented publicly |
| Optimised for | typed knowledge graph extraction (ECL pipeline) over RAG | LLM-as-OS paradigm; agent autonomy in managing its own memory |
| Anti-fit | no anti-fit explicitly stated | no anti-fit explicitly stated |
Taxonomy
| Axis | Cognee | Letta / MemGPT |
|---|---|---|
| storage | graph | kv |
| retrieval | graph-traversal | similarity |
| persistence | long-term | long-term |
| update | extraction | agent-controlled |
| unit | fact | episode |
| governance | inspectable | inspectable |
| conflict | llm-arbitrate | llm-arbitrate |
Pros & cons
Cognee
Pros: Pipeline-as-code ECL (extract/cognify/load) makes the memory build path inspectable and replayable; fully OSS.
Cons: Smaller community than Mem0/Letta; more end-to-end engineering required to deploy.
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.