Hindsight (Vectorize) vs Letta / MemGPT
Hindsight (Vectorize) 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
| Hindsight (Vectorize) | Letta / MemGPT | |
|---|---|---|
| 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, Offline Capable | 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.
| Hindsight (Vectorize) | Letta / MemGPT | |
|---|---|---|
| Use cases | Long Running Session, Memory Augmented Chat, Offline Capable | Long Running Session, Memory Augmented Chat, Multi Agent Coordination |
| Type | Vector + reflection / summarisation | Tiered: core / recall / archival |
| Created | 2025-10 | 2023-10 |
| Latest release | v1.1.0 (2026-10-21) | 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 | MIT | Apache-2.0 |
| GitHub | vectorize-io/hindsight — ~44.6k 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 | OSS self-host free (MIT); Hindsight Cloud pay-as-you-go credits ($5 free credit on GitHub sign-up); Enterprise single-tenant custom | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom |
| Funding | total $3.6M (Seed led by True Ventures, 2024-04) | $10M total $70M val Seed · 2024-09 |
| Backend storage | custom | Postgres + pgvector |
| Deployment | Both — managed Hindsight Cloud; Enterprise single-tenant (Vectorize cloud, customer AWS/GCP/Azure, or on-prem); self-host via Docker, Helm or pip (Postgres + pgvector) | Both |
| Multi-tenancy | namespace | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation |
| MCP | native (first-party) — Hindsight MCP | native (first-party) — Letta serves & consumes MCP |
| Optimised for | retrieval engineering + reflection / summarisation as service | LLM-as-OS paradigm; agent autonomy in managing its own memory |
| Anti-fit | searched not found | no anti-fit explicitly stated |
At a glance
| Hindsight (Vectorize) | Letta / MemGPT | |
|---|---|---|
| Section | Dedicated memory layers | Dedicated memory layers |
| Tier | T1 | T1 |
| Type | Vector + reflection / summarisation | Tiered: core / recall / archival |
| Created | 2025-10 | 2023-10 |
| Latest release | v1.1.0 (2026-10-21) | 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 | MIT | Apache-2.0 |
| GitHub | vectorize-io/hindsight — ~44.6k 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 | OSS self-host free (MIT); Hindsight Cloud pay-as-you-go credits ($5 free credit on GitHub sign-up); Enterprise single-tenant custom | Free $0 (3 agents); Pro $20/mo; Developer pay-as-you-go ($0.10/active agent/mo); Teams; Enterprise custom |
| Funding | total $3.6M (Seed led by True Ventures, 2024-04) | $10M total $70M val Seed · 2024-09 |
| Backend storage | custom | Postgres + pgvector |
| Deployment | Both — managed Hindsight Cloud; Enterprise single-tenant (Vectorize cloud, customer AWS/GCP/Azure, or on-prem); self-host via Docker, Helm or pip (Postgres + pgvector) | Both |
| API surface | REST, SDK: Python, TS | REST, SDK: Python, TS |
| Embedding | multiple supported | multiple supported |
| Multi-tenancy | namespace | Logical namespace per org/project/agent; self-hosted Docker option for hard tenant isolation |
| MCP | native (first-party) — Hindsight MCP | native (first-party) — Letta serves & consumes MCP |
| A2A | not documented publicly | not documented publicly |
| OpenTelemetry | not documented publicly | not documented publicly |
| Optimised for | retrieval engineering + reflection / summarisation as service | LLM-as-OS paradigm; agent autonomy in managing its own memory |
| Anti-fit | searched not found | no anti-fit explicitly stated |
Taxonomy
| Axis | Hindsight (Vectorize) | Letta / MemGPT |
|---|---|---|
| storage | vector | kv |
| retrieval | similarity | similarity |
| persistence | long-term | long-term |
| update | consolidation | agent-controlled |
| unit | episode | episode |
| governance | opaque | inspectable |
| conflict | none | llm-arbitrate |
Pros & cons
Hindsight (Vectorize)
Pros: Built on Vectorize's RAG pipeline expertise — retrieval tuning is a first-class concern; strong Pydantic-AI integration story.
Cons: Newer entrant with smaller adoption; positioning straddles RAG and memory which can muddy the value prop vs pure-play layers.
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.