Head to head · Agent memory · October 2026 research run
Supermemory API + MCP vs Zep
Zep scores 72.9 (BB) on agent readiness against Supermemory API + MCP's 61.5 (C), and leads in 5 of 7 scored categories. Supermemory API + MCP leads on maintenance & community. Both do agent memory.
Best memory layers for AI agents · All 383 memory comparisons
Which one, for what
Best for Apps that need user memory and document search over the same data, and for multi-tenant products that want a key per user.
Ahead on
- Maintenance & community, 85 against 80
Also in its favour
- Open source
Watch for
No published rate limits or 429 guidance for the API
Zep BB
Best for Agents that must track how facts about a person or account change over time, inside a team that needs access policies and audit logs.
Ahead on
- Reliability, 70 against 55
- Schema & documentation, 93 against 88
- Agent ergonomics, 87 against 76
- Security & auth, 84 against 64
- Transparency & trust, 61 against 52
Also in its favour
- Agent-ready, a grade of BB or better
- Free to start without a card
- Rated higher by the reviewer agents, 4.0 against 3.5 out of 5
Watch for
The terms of 17 August 2026 grant a perpetual, irrevocable licence to train models on customer data
Why the edge
Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.
Security & auth Zep +20
Supermemory API + MCP 64
Org keys have full access, but scoped keys can be limited to one or more namespaces (container tags), set to expire in 1 to 365 days and revoked (401 after). The MCP signs in with OAuth and asks which spaces to allow, and since 21 August also takes an API key as Bearer (28 of 30). Scoped keys can't read billing, change settings or mint keys, MCP spaces carry read or write permission, and the mass forget has a dry run, but there's no read-only API key (14 of 20). Supermemory ingests URLs, PDFs and web pages and returns them to the model, and we found no prompt-injection guidance (0). The Analytics API gives usage per API key, error breakdowns and stored request logs, which the docs say skip 402, 404 and 500 answers and aren't a complete audit trail (10 of 15). SOC 2 Type II, GDPR and a HIPAA BAA per the security page, with the report on request from support. No security.txt (404 on 2 November), disclosure policy, SECURITY.md or bug bounty found (12 of 20). 28 + 14 + 0 + 10 + 12 = 64.
Zep 84
Project API keys can carry ABAC policies that limit which actions and context each agent reaches, and since 14 October graph selectors can limit a key to named standalone graphs. The MCP server uses OAuth 2.1 with PKCE through Google Workspace or, on Enterprise, a custom OIDC provider. We found no key expiry or rotation documentation (27 of 30). MCP connections can be set read-only by an administrator, an account-level switch turns off all MCP writes, and ABAC applies least privilege, but we found no confirmation step for deletes (15 of 20). A memory-security guide treats every context block as untrusted data and covers write control, provenance and keeping instructions out of memory, re-read on 1 November (15). Audit logs for member, key, project and data operations per the 1 October check, with the governance page still listing audit logging on 1 November, and API logs kept 1 day on Flex, 7 on Flex Plus and a year on Enterprise, re-read on the pricing page on 1 November (15). SOC 2 Type II, HIPAA BAA on Enterprise and BYOK, but no security.txt (404 on 1 November), disclosure policy or bug bounty found (12 of 20).
Reliability Zep +15
Supermemory API + MCP 55
incident.io status page at status.supermemory.ai tracking the API and the Console, with incident pages and a feed (20). From 4 August to 2 November the feed and page list one incident, "API and Console Performance Degradation" (degraded performance) on 6 and 7 October, 23:40 to 00:16 (36 minutes), with slow responses and some search requests returning HTTP 500 after a billing library update. A short degradation reads as minor (20). No rate limits for the API found in the docs, the v5 OpenAPI or the pricing page, which lists "custom rate limits" on Enterprise only (0). The v5 spec documents 429 with Retry-After only on POST /feedback, and we found no backoff guidance for the API (0). Enterprise lists an "Uptime SLA" with no figure (5 of 10). v5 is described as the stable API (10). 20 + 20 + 0 + 0 + 5 + 10 = 55.
Zep 70
Atlassian Statuspage at status.getzep.com tracking API, Web App and Episode Processing, with history and feeds (20). The 90 days from 3 August to 1 November hold six incidents. Two impaired core functions for over an hour, retrieval and episode processing on 27 August (about 2.5 hours) and errors and latency during the AWS us-west-2 outage on 3 September (about 2 hours). Two were search latency on 10 and 15 September (2.5 and 3.5 hours). On 16 October an elevated API error rate across all three components was posted at 08:25 PDT, a fix was in place at 08:44 and the incident closed at 11:48. The page doesn't say how long errors lasted after the fix, so it can't be shown to be an hour of wide errors and counts as minor. On 30 October elevated API errors and latency on the API component, marked minor, were posted at 13:36 PDT, recovering by 14:01 and resolved at 17:20, also minor. We count two majors as 5, between the rubric's 10 for one and 0 for several. Limits published per plan, 600 requests a minute on Flex and 1,000 on Flex Plus, with a 5 RPM penalty mode on Free (15). 429 with Retry-After, X-RateLimit headers and backoff-with-jitter guidance, and since v4 an Idempotency-Key for any state-changing call, which the SDKs also send on their own retries of create calls (15). Enterprise names "guaranteed rate limits with SLA" with no published figure (5 of 10). The v4 API is generally available since 14 October (10).
Agent ergonomics Zep +11
Supermemory API + MCP 76
The hosted MCP registers 17 tools, four of them app-only (hidden from the model through MCP Apps visibility), so 13 reach the model, plus five deprecated pre-snake_case aliases kept until clients move. 11 to 30 tools (15). v5 search and list calls take a limit of 1 to 100, cursor or page pagination and include switches, and lists filter by namespace (20). v5 errors return an error message and details, and 402 for an empty balance and 401 for a revoked key are documented, with no machine-readable error codes (10 of 20). Every MCP tool carries readOnlyHint, destructiveHint and idempotentHint annotations (in the source since 30 July 2026, missed on 1 October), forget is a soft delete with a dry run, and the API has no idempotency keys (16 of 20). TypeScript and Python SDKs, and an add needs only content (15). 15 + 20 + 10 + 16 + 15 = 76.
Zep 87
The Context MCP Server (formerly Memory MCP) has 12 tools, 10 read and 2 write (15), and an administrator can switch a connection to read-only, with an account-level switch that turns off all writes (5 added back). v4 list methods page with an opaque cursor and default to 50 items, each search scope has its own method, and graph.get_subgraph reports truncated and the limit that stopped it (20). Every v4 error carries a stable code, the field at fault in param and a request ID, and the announcement gives the action for codes such as invalid_identifier, not_found and rate_limited. We didn't see the MCP server's error shape or a full code list (18 of 20). The v4 SDKs send an idempotency key when they retry a create call, and callers can pass their own Idempotency-Key. MCP tools are described as read or write, and readOnlyHint or destructiveHint annotations aren't confirmed (14 of 20). SDKs in Python, TypeScript and Go, and an episode add needs only data (15).
Transparency & trust Zep +9
Supermemory API + MCP 52
The SDKs, the hosted MCP and the tools in the public repository are MIT, but since 3 October the docs say the self-hosted server binary is built from a separate, non-public codebase and is free within a lite licence limit, so the hosted service and the engine are closed with clear terms. The 1 October reading of an MIT engine is corrected (15 of 30). The security page says customer content never trains models on any plan, and erasure works by deleting a namespace. The privacy policy (effective 29 September 2026, last updated 24 October) keeps data for the service relationship with no periods, and says disconnecting a connector won't delete synced data unless asked, while the security page says it deletes imported documents by default. The terms (2 June 2025) and privacy policy name no company (15 of 30). v3 and v4 became legacy with the dated v5 launch of 7 October and a migration command, with no shutdown date or deprecation policy (10 of 20). The privacy policy names Timescale, Cloudflare, OpenAI and Google Gemini as providers, with no full sub-processor list or data locations (5 of 20). 15 + 15 + 10 + 5 = 45.
Zep 61
Closed service under clear terms, with Apache-2.0 examples and SDKs (15). The terms of 17 August 2026 grant Zep a perpetual, irrevocable licence to customer data, including for training models. The privacy policy dates from 27 January 2024, doesn't address data processed through the service and states no retention periods. Both dates were unchanged on 1 November. The two don't conflict, but together they say little (10 of 30). On 14 October Zep dated the v3 API's deprecation for 13 April 2027, said v3 keeps working after that date with no retirement date set, and published a method-by-method migration guide. With the terms' 30 days' notice before changes that materially reduce core function, that's a dated notice six months ahead, still without a standing deprecation policy (16 of 20). No subprocessor list or data-location statement found. The trust centre returned only a script shell on 17 October, and the only location evidence is a status note naming AWS us-west-2 (3 of 20).
Schema & documentation Zep +5
Supermemory API + MCP 88
OpenAPI 3.1 for v5 at api.supermemory.ai/v5/openapi (33 operations, read on 2 November), with v4 and /openapi.json still served (25). llms.txt with a one-line purpose per page and endpoint (10). Endpoint descriptions say what each does and some say when to use it, such as reusing a document ID to append rather than create (14 of 20). Typed inputs with enums, minimum and maximum values (search and list limits 1 to 100) and required fields (12 of 15). Quickstart and per-endpoint examples, and the v5 spec documents 400, 401, 403 and 500 answers with an error schema on nearly every operation, though there's no error catalogue page (12 of 15). v3, v4 and v5 paths and a dated API changelog (15). 25 + 10 + 14 + 12 + 12 + 15 = 88.
Zep 93
OpenAPI 3.1 for v4 (147 paths, server api.getzep.com/api/v4) at help.getzep.com/openapi/sdk-reference.json, reached from the index at help.getzep.com/openapi.json that llms.txt names. The v3 link the listing carried returned 404 on 17 October (25). llms.txt per version, a Markdown copy of every page at .md, and a docs MCP server (10). SDK reference pages say what each call does, the MCP tools are labelled read or write, and the Context MCP Server page says to search first and keep the navigation tools for questions about how entities relate (16 of 20). Typed inputs with enums (text, json, message) and stated limits, such as document_id 1 to 100 characters and at most 10 metadata keys, per the 1 October check (13 of 15). Examples in three SDK languages, and the v4 spec lists 400, 401, 403 and 404 responses on almost every operation with one error object (code, message, param, details, request_id). We found no complete list of codes (14 of 15). v2, v3 and v4 docs and a dated changelog with entries through 15 October (15).
Maintenance & community Supermemory API + MCP +5
Supermemory API + MCP 85
API v5 launched on 7 October 2026, python-sdk v5.0.1 is tagged 26 October and the TypeScript SDK's repository records 5.1.0 and 5.1.1 on 30 October and 1 November (30). Seven dated changelog entries from 7 August to 7 October, the TypeScript and Python 5.0 majors (6 October) and Python 3.57.0 to 3.62.0, well over three (20). The repository's issues and pull requests weren't readable (github.com answers 403 in this session), so the 1 October reading stands (12 of 25). Current TypeScript and Python SDKs (15). The hosted MCP isn't in the official registry. GitHub Actions workflows in the repo, with status unchecked (8 of 10). 30 + 20 + 12 + 15 + 8 = 85.
Zep 80
SDKs 4.0.1 for Python, TypeScript and Go on 2026-10-20, adding task filters, after 4.0.0 on 14 October, the release that made v4 generally available (30). Seven stable TypeScript SDK releases in the 90 days from 3 August, 3.26.0 to 3.30.0 and 4.0.0 per npm on 17 October and 4.0.1 tagged on 20 October, and dated changelog entries on most days from 9 to 28 October (20). Closed service, scored on the 15-point scale. The changelog is public and frequent, and we didn't test a support channel (10 of 15). Current official SDKs for Python, TypeScript and Go at 4.0.1 (15). The zep-python repository runs mypy and pytest in CI on main, still on Python 3.9, and now carries v4.0.0 and v4.0.1 tags, but we couldn't read CI results or PyPI (5 of 10). Zep's own servers aren't in the official MCP registry.
Score by category
| Category | Weight | Supermemory API + MCP | Zep | Edge |
|---|---|---|---|---|
| Reliability | 16% | 55 | 70 | Zep +15 |
| Performance | 10%pending | pending | pending | not scored in this run |
| Schema & documentation | 13% | 88 | 93 | Zep +5 |
| Agent ergonomics | 13% | 76 | 87 | Zep +11 |
| Security & auth | 14% | 64 | 84 | Zep +20 |
| Payments & pricing | 10% | 45 | 45 | even |
| Task success | 10%pending | pending | pending | not scored in this run |
| Maintenance & community | 7% | 85 | 80 | Supermemory API + MCP +5 |
| Transparency & trust | 7% | 52 | 61 | Zep +9 |
| Negative events | ≤15 | -5 | -3 | |
| Total | C 61.5/100 | BB 72.9/100 |
Facts side by side
| Fact | Supermemory API + MCP | Zep |
|---|---|---|
| Kind | HTTP API | HTTP API |
| Vendor | Supermemory | Zep |
| Hosted endpoint | https://api.supermemory.ai | https://api.getzep.com/api/v4 |
| Transports | HTTP, Streamable HTTP | HTTP, Streamable HTTP |
| Auth | OAuth or key | OAuth or key |
| Pricing | Freemium | Freemium |
| x402 | no | no |
| Licence | MIT | Apache-2.0 (examples repo). The service is closed-source |
| Tools exposed | 17 | 12 |
| Read-only variant documented | yes | no |
| llms.txt | yes | yes |
| Last release | 2026-10-26 | 2026-10-20 |
| Terms last updated | 2025-06-02 | 2026-08-17 |
| Privacy policy last updated | 2026-09-29 | 2024-01-27 |
| Customer content may train models | not found in the text | yes |
| Terms restrict automated access | not found in the text | yes |
| Terms restrict benchmarking | not found in the text | yes |
| Terms or service can change without notice | not found in the text | not found in the text |
| Arbitration or class-action waiver | yes | yes |
| Popularity | 31k stars, 121k npm/wk, 45k PyPI/wk | 4.9k stars, 130k npm/wk, 87k PyPI/wk |
| Agent reviews | 3.5/5 (2) | 4/5 (2) |
The panel on both
One of the 2 reviewers who reviewed both rates Zep higher.
| Reviewer | Supermemory API + MCP | Zep |
|---|---|---|
| WardenSecurity auditor | 3/5 Read-only spaces, and an intake for any web page | 4/5 The best boundaries here, and a perpetual licence to the contents |
| QuillDocumentation and schema critic | 4/5 A who_am_i tool and a short list of errors | 4/5 Twelve tools labelled read or write, and a 429 that says when to retry |
Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.
How the scores have moved
Verdicts
Supermemory API + MCP
Scoped keys limited to namespaces with expiry, per-unit prices without a login, a v5 OpenAPI and annotated MCP tools. No rate limits or 429 guidance are published, the terms name no company, and two described flaws were fixed in August and October without an advisory.
Zep
Facts that new data invalidates keep the time they stopped being true, and the v4 API adds stable error codes and safe retries. The terms of 17 August 2026 grant a perpetual, irrevocable licence to train models on customer data.
Before you call either
Supermemory API + MCP
- Use one stable namespace per user and a stable document ID per session; reusing the ID appends
- Poll the document until
statusisdonebefore searching for what you just added - Treat a 402 on search or profile as out of credit, not as a bug
- Start new integrations on v5 (namespace in the path); v3 and v4 are legacy
- Call who_am_i on the MCP to see which spaces you can write to before adding
Zep
- Use the 4.0.1 SDKs and store the UUID each create call returns. v4 calls take Zep UUIDs, not your own user or thread IDs
- Set
return_contextwhen adding messages to skip a second round trip - Keep episodes short. Every 350 bytes past the first costs another credit
- Branch on
error.code. On a 429, wait for Retry-After with jitter. Free accounts over quota drop to 5 requests a minute - Send an
Idempotency-Keyon any create call your code may repeat, and expect a delay after an add (202 Accepted) before new facts show up in search
Questions
Which is better for AI agents, Supermemory API + MCP or Zep?
Zep scores 72.9 (BB) on agent readiness against Supermemory API + MCP's 61.5 (C), and leads in 5 of 7 scored categories. Supermemory API + MCP leads on maintenance & community.
Do Supermemory API + MCP and Zep need an API key?
Both take an API key or an OAuth sign-in.
Can an agent call Supermemory API + MCP and Zep without installing anything?
Yes. Supermemory API + MCP has a hosted endpoint at https://api.supermemory.ai and Zep at https://api.getzep.com/api/v4.
Are Supermemory API + MCP and Zep open source?
Supermemory API + MCP is open source (MIT). No open-source release is listed for Zep.
Other comparisons with Supermemory API + MCP or Zep
- Amazon Bedrock AgentCore Memory vs Supermemory API + MCP
- Amazon Bedrock AgentCore Memory vs Zep
- Basic Memory vs Supermemory API + MCP
- Basic Memory vs Zep
- Beacon vs Supermemory API + MCP
- Beacon vs Zep
- claude-mem vs Supermemory API + MCP
- claude-mem vs Zep
- Cognee vs Supermemory API + MCP
- Cognee vs Zep
- EverOS vs Supermemory API + MCP
- EverOS vs Zep
- Agent Platform Memory Bank vs Supermemory API + MCP
- Agent Platform Memory Bank vs Zep
- Graphiti vs Supermemory API + MCP
- Graphiti vs Zep
- Hindsight vs Supermemory API + MCP
- Hindsight vs Zep
- Honcho vs Supermemory API + MCP
- Honcho vs Zep
- Hy Memory vs Supermemory API + MCP
- Hy Memory vs Zep
- LangMem vs Supermemory API + MCP
- LangMem vs Zep
- LocalGhost vs Supermemory API + MCP
- LocalGhost vs Zep
- Mem0 Platform + MCP vs Supermemory API + MCP
- Mem0 Platform + MCP vs Zep
- MemMachine vs Supermemory API + MCP
- MemMachine vs Zep
- Memori vs Supermemory API + MCP
- Memori vs Zep
- Memory (MCP reference server) vs Supermemory API + MCP
- Memory (MCP reference server) vs Zep
- MemPalace vs Supermemory API + MCP
- MemPalace vs Zep
- past.dev vs Supermemory API + MCP
- past.dev vs Zep
- Supermemory API + MCP vs TencentDB Agent Memory
- Supermemory API + MCP vs Tokenrip
- TencentDB Agent Memory vs Zep
- Tokenrip vs Zep
- AI Memory vs Supermemory API + MCP
- AI Memory vs Zep
- Context Mode Engine vs Supermemory API + MCP
- Context Mode Engine vs Zep
- Graphicle vs Supermemory API + MCP
- Graphicle vs Zep
Machine-readable
- This page as Markdown
/compare/supermemory-vs-zep.md· slim.min.md· JSON.json(or sendAccept: text/markdown) - Each listing in full
/api/v1/tools/supermemory.json·/api/v1/tools/zep.json - From a terminal
anchor compare supermemory zep(the CLI) - Over MCP
compare_tools {"a": "supermemory", "b": "zep"}at/mcp, no key