Context graphs for enterprise agents.
A shared memory for people across AIs.
Zep calls itself the enterprise context layer for reliable agents: Context Lake turns business data into governed context graphs, built on Graphiti, its open-source temporal graph framework. Kemory overlaps on agent memory over MCP, but it starts from the conversations people have with AIs, and from sharing them safely.
Last verified 7 November 2026 against primary sources
Two good answers.
Different jobs.
- You are building enterprise agents that need governed, graph-structured context over business data.
- Temporal reasoning (what was true, and when) and provenance back to source data are central to your application.
- You want an open-source temporal knowledge-graph framework on your own graph database.
- You want your own and your team’s chats in six AIs captured automatically, then recalled or continued in another AI.
- You need to share memory with people across organisations, with accept-to-receive, revocation and audited reads.
- You want published plan prices that start free, with governance on every plan.
See each one
doing its job.
Screenshots taken 7 November 2026. Zep / Graphiti is a trademark of its owner, shown here to identify the product.
At a glance
The same rows on every comparison, so you can read one page against another.
*89.6% end-to-end answer accuracy on LongMemEval-S, open-weight Llama-3.3-70B reader, six runs. See the benchmarks or the full method on docs.sekondbrain.ai. A dash means not documented, not proven absent.
The same job,
done two ways.
Graphiti tracks what was true and when, and links every fact back to its source episode, a strong fit for questions about change over time. Context Lake brings structured and unstructured business data into one graph, and Zep integrates with a wide range of agent frameworks, from Google ADK and LangGraph to the Vercel AI SDK.
The numbers,
and how each was run.
These rows use different datasets, splits, readers and retrieval budgets, so they cannot be compared directly. We publish the method, every run including the lower ones, and use an open-weight reader rather than a frontier model.
| Product | Benchmark | Score | What it measures | Setup | Reader model | Runs |
|---|---|---|---|---|---|---|
| Kemory | LongMemEval‑S | 89.6% | End-to-end answer accuracy | 500 questions, oracle split | Llama-3.3-70B (open-weight) | Six, with variance |
| Zep / Graphiti | DMR, LongMemEval (Zep paper, Jan 2025) | Not verified | Self-reported | Not verified | Not verified | Not verified |
What changed at Zep / Graphiti
Products move quickly. These are the dated changes this comparison has been checked against.
- 14 October 2026Zep’s integration notes describe it as “a paid product (no meaningful free tier)”.
- 20 October 2026Graphiti 0.31.2 released.
- 2 November 2026zep-cloud SDK 4.1.0 released.
- 4 November 2026getzep/zep rewritten around Context Lake and Agent Skills; legacy Community Edition code removed.
Specialists do one job well.
Kemory holds the thread across them all.
Every year brings better models and you will keep trying them. Kemory sits above the models rather than inside any one of them, so what you have worked out goes with you.
Tell one AI something once, and every AI you use just knows it. Start a thought in one and carry on in another, mid-sentence.
Every memory is permissioned and every change is provable. We do not train on your data. Delete anything, and prove it is gone.
Compaction runs on an open-weight model, so the frontier model only ever sees the compacted version — not the pile.
Already using Zep / Graphiti? You do not have to replace it to try Kemory. Kemory connects to the AIs you use over MCP and Kora for Chrome.
Choosing Zep / Graphiti instead?
Good — it fits you better right now.
Come back to Kemory when cross-AI, governed memory is what you need.
Five questions,
Kemory vs Zep / Graphiti.
- What is the difference between Kemory and Zep?
- Zep is an enterprise context layer. It turns business data into governed context graphs for agents, using Graphiti’s temporal graph. Kemory is a cross-AI memory for people and teams. It captures chats from the major AIs, recalls them over MCP, and governs sharing with consent and audit.
- Do Kemory and Zep support MCP?
- Yes. Zep says agents connect through MCP or its Python, TypeScript and Go SDKs, and Graphiti ships its own MCP server. Kemory offers a one-click connector in Claude, ChatGPT and four more AIs, and MCP recall in about twenty more clients.
- How are Zep and Kemory priced?
- Zep’s own integration notes describe it as a paid product with no meaningful free tier; we could not verify plan prices on 7 November 2026. Kemory is free for 2,000 memories and two AIs, with Pro at $19.99 a month, Team at $30 a seat and Business at $55 a seat (10-seat minimum).
- Can I still self-host Zep?
- The legacy Zep Community Edition code was removed from the getzep/zep repository on 4 November 2026. Graphiti, the open-source framework under Zep, is still Apache 2.0 and runs on your own Neo4j, FalkorDB or Amazon Neptune.
- Are Zep’s benchmark results comparable with Kemory’s?
- Not directly. Zep’s January 2025 paper reports DMR and LongMemEval results, and we have not re-verified their split or reader model. Kemory reports 89.6% end-to-end answer accuracy on the LongMemEval-S oracle split with an open-weight Llama-3.3-70B reader, across six runs. Different splits, readers and retrieval budgets give different numbers.
More comparisons
Credit where it’s due. The real difference, either way.