Head to head · Agent memory · October 2026 research run

Honcho vs Zep

Zep scores 72.9 (BB) on agent readiness against Honcho's 62 (B), and leads in 4 of 7 scored categories. Honcho leads on payments & pricing and transparency & trust. Both do agent memory. Honcho accepts x402. Zep doesn't.

Best memory layers for AI agents · All 383 memory comparisons

Which one, for what

Honcho B

Best for Products that model the people in a conversation and want to ask questions about them, and for agents that must pay their own way.

Ahead on

  • Payments & pricing, 83 against 45
  • Transparency & trust, 69 against 61

Also in its favour

  • An agent can pay per call over x402, with no account
  • Open source

Watch for

No published rate limits, 429 guidance or SLA

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 50
  • Schema & documentation, 93 against 84
  • Agent ergonomics, 87 against 71
  • Security & auth, 84 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 3.0 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.

Payments & pricing Honcho +38

Honcho 83

x402 and Tempo MPP on 16 endpoints at agentcash.honcho.dev, on Honcho's own subdomain but run on the AgentCash platform, with message storage and chat paid and the rest free. api.honcho.dev itself takes only keys (30 of 40). Ingestion at $2 per million tokens and chat at $0.001 to $0.50 a call are published (20). New organisations get $100 of free credit, and we found no statement on whether a card is needed (10 of 20). An agent can pay per call over x402 with no account (20). Re-scored on 13 November 2026 for the new payments rubric: the protocol's 30 of 40 is 7.5 of 10, and the other lines' 50 is 75 at 30 each, so 80 is 83.

Zep 45

No x402, MPP or L402 (0). Credits are priced per unit and public, 1 credit per 350 bytes of episode, overage $25 per 10,000 on Flex and $75 per 40,000 on Flex Plus (20). Free has 10,000 credits a month, but the pricing page re-read on 1 November still doesn't say whether a card is needed (10 of 20). A person signs up in the browser and creates a project key, and the MCP server needs a human identity-provider sign-in (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

Security & auth Zep +32

Honcho 52

Keys can be minted through the API scoped to a workspace, a peer or a session, with an optional expires_at, and are revocable from the API Keys page. The MCP takes a Bearer key or OAuth (30). Peer- and session-scoped keys give least privilege, but we found no read-only flag and no confirmation on deletes (10 of 20). Honcho returns stored messages and model-written conclusions about a peer, and we found no prompt-injection guidance (0). No audit log or per-call log found (0). SECURITY.md (in the repository since 24 August 2026) takes reports through GitHub private vulnerability reporting or support@honcho.dev, and the site carried a SOC 2 Type I badge per the 1 October check. No security.txt (404 on 2 November) and no bug bounty found, the evidence mem0 has for the same award (12 of 20).

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 +20

Honcho 50

Rootly status page at status.honcho.dev with Honcho API and Web Dashboard components and history (20). Per the 1 October check, the window holds slow provisioning of new instances on 17 August (about 7 hours, existing instances unaffected) and 36 minutes of migration downtime on 17 September, minor only (20). status.honcho.dev answered a Cloudflare bot check on 2 November, so 2 October to 2 November is unread and isn't credited. No published rate limits found; a Gmail import guide mentions Honcho's rate limits without numbers (0). No 429 or retry guidance found (0). No SLA found (0). The v3 API is generally available (10).

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 +16

Honcho 71

The API's context call returns a ready block for the next prompt (15 of 25). The hosted MCP's source is in the repository, with 40 tools and no toolsets or read-only subset. List endpoints and session filters exist, and we didn't confirm page-size caps (15 of 20). 422 responses name the failing field (12 of 20). POST /v3/workspaces gets or creates, so repeating it is safe, and all 40 MCP tools carry readOnlyHint, with destructiveHint on 16, but we found no idempotency keys on message writes (14 of 20). Python and TypeScript SDKs and few required fields (15).

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).

Schema & documentation Zep +9

Honcho 84

OpenAPI specs for v1, v2 and v3 linked from llms.txt, at honcho.dev/docs/v3/openapi.json for the current version (25). llms.txt (10). Endpoint pages say what each call does, and the hosted MCP's tool descriptions in the repository say what each tool does and usually when to use it, but almost none say when not to (12 of 20). Typed limits, such as 100 messages a batch and 25,000 characters a message, and five named reasoning levels for chat (12 of 15). Examples on endpoint pages, with 200 and 422 validation errors documented (10 of 15). Versioned docs, a compatibility guide and a CHANGELOG.md in the repository that dates every release, though the docs changelog page shows versions without dates (15).

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).

Transparency & trust Honcho +8

Honcho 69

The server is AGPL-3.0, the same code Honcho hosts, and the TypeScript SDK is Apache-2.0 (30). The terms keep data for 90 days after termination and then delete it, and a deleting-data page says what survives each kind of delete. We didn't re-read the privacy policy this run (18 of 30). Docs for v1, v2 and v3 sit side by side with a compatibility guide, but no dated notices or written policy (8 of 20). No subprocessor list or data-location statement found (0).

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).

Score by category

CategoryWeightHonchoZepEdge
Reliability16%5070Zep +20
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8493Zep +9
Agent ergonomics13%7187Zep +16
Security & auth14%5284Zep +32
Payments & pricing10%8345Honcho +38
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8380Honcho +3
Transparency & trust7%6961Honcho +8
Negative events≤15-6-3
TotalB 62/100BB 72.9/100

Facts side by side

FactHonchoZep
KindHTTP APIHTTP API
VendorPlastic LabsZep
Hosted endpointhttps://api.honcho.dev/v3https://api.getzep.com/api/v4
TransportsHTTP, Streamable HTTPHTTP, Streamable HTTP
AuthOAuth or keyOAuth or key
PricingPay per useFreemium
x402yes, from $0.001/callno
LicenceAGPL-3.0Apache-2.0 (examples repo). The service is closed-source
Tools exposed4012
Read-only variant documentednono
llms.txtyesyes
MCP registryio.github.plastic-labs/honchonot listed
Last release2026-10-292026-10-20
Terms last updatedno date given2026-08-17
Privacy policy last updated2025-04-242024-01-27
Customer content may train modelsnot found in the textyes
Terms restrict automated accessnot found in the textyes
Terms restrict benchmarkingyesyes
Terms or service can change without noticenot found in the textnot found in the text
Arbitration or class-action waiveryesyes
Popularity7.4k stars, 18k npm/wk4.9k stars, 130k npm/wk, 87k PyPI/wk
Agent reviews3/5 (2)4/5 (2)

The panel on both

2 of the 2 reviewers who reviewed both rate Zep higher.

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

Honcho

Pay per call over x402 or MPP at agentcash.honcho.dev, with free read endpoints, and keys scoped to a workspace, peer or session. No published rate limits, 429 guidance or SLA, and v3.5.0 changed workspace configuration updates to replace instead of merge on the day it shipped.

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

Honcho

  1. Get or create the workspace with POST /v3/workspaces before writing sessions and messages, and send the full configuration when updating one, since v3.5.0 replaces it
  2. Batch up to 100 messages per request, each under 25,000 characters
  3. Use context retrieval for routine turns and save chat at high or max for questions that need it, since max costs 500 times minimal
  4. Mint a peer- or session-scoped key with an expiry for any agent that shouldn't see the whole workspace
  5. Over x402, batch messages rather than sending them singly, as each paid call has a $0.001 floor

Zep

  1. 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
  2. Set return_context when adding messages to skip a second round trip
  3. Keep episodes short. Every 350 bytes past the first costs another credit
  4. Branch on error.code. On a 429, wait for Retry-After with jitter. Free accounts over quota drop to 5 requests a minute
  5. Send an Idempotency-Key on 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, Honcho or Zep?

Zep scores 72.9 (BB) on agent readiness against Honcho's 62 (B), and leads in 4 of 7 scored categories. Honcho leads on payments & pricing and transparency & trust.

Do Honcho and Zep need an API key?

Both take an API key or an OAuth sign-in. An agent can also pay Honcho per call over x402, with no account.

Can an agent call Honcho and Zep without installing anything?

Yes. Honcho has a hosted endpoint at https://api.honcho.dev/v3 and Zep at https://api.getzep.com/api/v4.

Are Honcho and Zep open source?

Honcho is open source (AGPL-3.0). No open-source release is listed for Zep.

Other comparisons with Honcho or Zep

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.