mem01 ( agent memory )
Self-hosted memory for AI agents that updates when the truth changes — so they don't keep acting on yesterday's facts.
$ docker compose up -d --build
bag of facts
Both active. Agent invents a cover story.
mem01 · recall (now)
Default: one truth for the agent.
mem01 · history / include_history
“Where did we live before SF?”
Previous facts stay in the DB and can be recalled on purpose — labeled so the agent never confuses past with present.
The problem
Users change their minds. Agents that remember everything end up remembering the past louder than the present.
Most memory layers ADD. Preferences flip, jobs change, cities move — the old fact still sits in the prompt.
When NY and SF both retrieve, the LLM covers the mess. Looks fine in a demo. Fails in production.
Token budgets fill with contradictions. Cost climbs. Correctness slides the other way.
Benchmarks
Self-run on LoCoMo-10 (1,540 questions) with the same open eval style as mem0's public harness. Same models for extract, answer, and judge: gpt-4o-mini.
v0 · baseline
31.7%
488/1540 correct
v0.2 · current
77.9%
1200/1540 correct · +46.2 pts
Full methodology, models, and caveats on the research page — including why this is not a copy of every published headline number.
Research detailsCompare
mem0 wins on cloud polish. Zep wins on graph depth. mem01 wins when you need correct beliefs under evolution, self-hosted, lean recall.
| Capability | mem01 | mem0 | Zep |
|---|---|---|---|
| Belief lifecycle (supersede / invalidate) | |||
| Stale facts retired on conflict | |||
| Recall without LLM by default | |||
| Token-budgeted context packing | |||
| Self-hosted first (your Postgres) | |||
| Light model — no full knowledge graph | |||
| Open-source MIT core |
● yes · — partial · × no · from public docs / positioning
vs mem0
They do well: Best-in-class ease and fact extraction.
Gap: After NYC → SF, search often returns both. Store stays dirty; agent covers it.
Why mem01: SUPERSEDE retires NYC. One active belief. Product suite: mem01 5/5 vs mem0 2/5 on conflict cases.
vs zep
They do well: Temporal graphs, enterprise packaging, multi-hop context.
Gap: Graph materialization trades simplicity and cost. Overkill for “two cities in the prompt.”
Why mem01: Belief ops + budgeted packing. Evolution correctness without a full temporal graph.
vs diy
They do well: You control embeddings and filters.
Gap: No extraction policy, no conflict model, no packer. Every agent reinvents hygiene.
Why mem01: remember / recall / correct / forget over Postgres. Lifecycle is the product.
Use cases
Not every product needs belief memory. These are the jobs where staleness, scopes, and self-host actually change the outcome.
Why here · Stack, monorepo tools, and “we use light mode now” change over a repo’s life. Append-only memory keeps old toolchains in the prompt.
How mem01 · Scope by user + project. SUPERSEDE when the project switches package manager, model, or conventions. Cursor / Claude / custom agents hit the same HTTP API.
Skip if · One-shot scripts with no shared state — skip a memory layer.
Why here · Location, job, plan tier, and preferences are single-slot truths. Stacked facts force the model to guess which one is current.
How mem01 · Default recall is active-only. Opt into history when the question is temporal (“before SF?”) without poisoning every turn.
Skip if · Pure multi-hop “who works with whom across orgs” — a temporal graph fits better.
Why here · Agent memory often includes code context, internal names, and PII. SaaS memory is a non-starter for many shops.
How mem01 · docker compose · Postgres + pgvector on your network. Same remember / recall shape as a managed API, under your keys.
Skip if · Zero-ops hobby demos that only need a cloud key and never leave the laptop.
Why here · One user, several agents: a coder, a support bot, a research runner. Shared user prefs should travel; agent scratch should not.
How mem01 · user / project / agent / session scopes. Share what’s meant to be shared; isolate the rest.
Skip if · A single chat window with no second agent or project boundary.
System
Send messages. One extraction pass classifies belief ops: ADD, UPDATE, SUPERSEDE, INVALIDATE, MERGE — then writes to Postgres + pgvector.
New truths replace old ones. Superseded and invalidated beliefs stay for audit, drop out of active recall.
Embed the query, filter by scope, apply conflict rules, pack to a token budget. No LLM on the hot path.
Product
ADD · UPDATE · SUPERSEDE · INVALIDATE · MERGE — not just store another string.
Vector search, scope filters, conflict rules, token packer. Fast hot path.
Budgeted packing keeps memory blocks tight so agents stay cheap.
Production store in dev and deploy. Neon or any Postgres with vectors.
docker compose up. Your network, your keys, your compliance boundary.
User, project, agent, session. Share what should be shared; isolate the rest.
API
HTTP + Python SDK. remember extracts beliefs; recall returns a budgeted context string. Also correct and forget when humans need to fix the model.
# remember
curl -s localhost:8080/v1/remember \
-H 'Content-Type: application/json' \
-d '{
"user_id": "user_1",
"messages": [
{"role": "user", "content": "I live in San Francisco."}
]
}'
# recall — no LLM on this path
curl -s localhost:8080/v1/recall \
-H 'Content-Type: application/json' \
-d '{
"user_id": "user_1",
"query": "Where does the user live?",
"max_memory_tokens": 800
}'FAQ
Same job — long-term agent memory — different model. mem01 is belief-centric: lifecycle ops, conflict filtering, and self-hosted Postgres as the default story.
No. v1 avoids full temporal graphs. You get validity windows and supersede chains without multi-hop search on every request.
Docker Compose: FastAPI + PostgreSQL with pgvector. Point OPENAI_API_KEY (or compatible) at extraction/embeddings. Data stays in your DB.
Yes by default. Recall is embed → search → filter → pack. Writes may call an LLM once per batch for extraction.
On conflict and staleness — location/job/preference flips — not every public benchmark. Product suite: mem01 5/5 vs mem0 2/5 when old values must not reappear.
No. Default recall is current-only so agents stay correct. For “before SF?” or medical audit, call recall with include_history=true, or POST /v1/history for a full timeline. Past beliefs stay labeled, not mixed in as active truth.
Get started
Clone the repo. Compose up. Call remember and recall.
Start on GitHub