Zep

Zep is a production agent-memory and context-engineering...

7.8/10 Useful

Active

Free 10k credits/month / Flex $125/month / Flex Plus $375/month / Enterprise custom

Best plan

Use the free 10k-credit path for prototypes, Flex when production...

Risk: Agent memory can contain sensitive user facts, business...

Try Zep free

Editorial · no paid placements

Should you use it?

Zep is a production agent-memory and context-engineering platform. Pick it when an AI product needs persistent memories, temporal context graphs, and context assembly across users, sessions, and business data. Compare Mem0 for managed/open-source memory choices, Qdrant or Weaviate for vector database control, and LangGraph or Mastra when the missing piece is agent orchestration.

Buy if AI agents that need persistent user and task memory
Pick Use the free 10k-credit path for prototypes, Flex when production agent memory needs 50k credits/month and basic limits, Flex Plus for 200k credits/month, observations, custom extraction, webhooks, analytics, and longer API logs, and Enterprise for SLA, BYOK, BYOC, HIPAA, audit logs, and custom rate limits
Skip if Teams that only need a vector database for document chunks

Plan guidance

What to buy

Best plan Use the free 10k-credit path for prototypes, Flex when production agent memory needs 50k credits/month and basic limits, Flex Plus for 200k credits/month, observations, custom extraction, webhooks, analytics, and longer API logs, and Enterprise for SLA, BYOK, BYOC, HIPAA, audit logs, and custom rate limits

Watch: Agent memory can contain sensitive user facts, business...

Price range Free 10k credits/month / Flex $125/month / Flex Plus $375/month / Enterprise custom

10,000 credits/month

Upgrade only if Not for teams that only need a vector database for document chunks

Agent memory can contain sensitive user facts, business...

Current pricing source: Zep pricing

Fit

Use it for this, skip it for that

Best for

AI agents that need persistent user and task memory
Teams building assistants that must remember facts across sessions
Products that need temporal context graphs instead of only vector retrieval
Enterprises that need cloud, BYOK, BYOC, HIPAA, and audit-log procurement options

Avoid if

Teams that only need a vector database for document chunks
Simple chatbots with no durable memory requirement
Buyers who cannot govern sensitive memory extraction and deletion
Teams that need a complete agent framework instead of a memory layer

Watch out: Agent memory can contain sensitive user facts, business context, retrieved documents, and inferred relationships, so buyers need consent, deletion, retention, BYOK/BYOC, and audit-log review before production.

Recent changes

Only what affects the decision

Jun 28, 2026
Free prototyping
Zep pricing describes a free prototyping path with 10,000 credits/month and limits
Zep pricing
Jun 28, 2026
Flex
Includes 50,000 credits/month, 600 requests/minute, 5 projects, 10 custom entity/edge types, 1-day API logs, 30-day rollover, and $25 per extra 10k credits
Zep pricing
Jun 28, 2026
Flex Plus
Includes 200,000 credits/month, 1,000 requests/minute, 10 projects, observations, custom extraction, webhooks, analytics, 7-day API logs, 60-day rollover, and $75 per extra 40k credits
Zep pricing

Alternatives

Best swaps

Hugging Face

Open AI collaboration hub for models, datasets, Spaces, inference endpoints, evaluations, and enterprise ML workflows.

Free hub access; Pro $9/mo; Team $20/user/mo; Enterprise from $50/user/mo; paid compute/storage · 9.3/10 LiteLLM

Open-source LLM gateway and Python SDK for one OpenAI-compatible interface across 100+ model providers, with routing, virtual ke

Free MIT core outside enterprise directory; Enterprise custom · 8.8/10 promptfoo

Open-source LLM evaluation, red teaming, vulnerability scanning, guardrails, model security, MCP proxy, code scanning, and enter

Community free / Enterprise custom / On-Premise custom · 8.8/10

Build comparison

Proof and score math Verified Jun 28

Proof

Why this recommendation is trusted

Evidence Zep official site

Source: Registered source
Freshness: Current
Confidence: High confidence
Verified: Jun 28, 2026
Review: Jul 12, 2026
Volatility: Volatile

High-volatility evidence needs frequent review.

Editorial score

Unweighted average of 4 axes · confidence high

Utility 8/10

How much real work it can do for a competent operator, end to end.
Value 7/10

What you get for the dollar relative to the closest alternative.
Moat 8/10

How hard it would be for a competitor to replicate the underlying advantage.
Longevity 8/10

How likely the product is to still be best-in-class 24 months out.

Verified facts

Best For Agent teams that need persistent memory, temporal context graphs, and automated context assembly from chat history and business data.
high Drifts 2026-06-28 Zep official site
Pricing Anchor Zep pricing lists a free 10k-credit prototyping path, Flex at $125/month or $1,250/year billed annually, Flex Plus at $375/month or $3,750/year billed annually, and Enterprise custom.
high Volatile 2026-06-28 Zep pricing
Watch Out For Agent memory can contain sensitive user facts, business context, retrieved documents, and inferred relationships, so buyers need consent, deletion, retention, BYOK/BYOC, and audit-log review before production.
high Drifts 2026-06-28 Zep docs concepts
Open Source Or Local Zep's Graphiti repository is Apache-2.0 licensed, but the Zep managed product is priced separately as a commercial agent-memory platform.
high Drifts 2026-06-28 Graphiti license
Usage Model Credits are consumed by Episode size: one credit up to 350 bytes, another credit for each additional 350 bytes or part, with memory, retrieval, storage, threads, users, and graph storage listed as unmetered.
high Volatile 2026-06-28 Zep pricing

Full review notes Long-form details, FAQ, and source history

Zep is an agent-memory and context-engineering platform. It is built for products where an agent needs to remember facts, relationships, and time-sensitive context across sessions instead of only retrieving documents from a vector index.

The plain-English fork: choose Zep when memory itself is the product risk. Choose a vector database when the problem is mostly document search.

System Verdict

Pick Zep when agents need persistent, scoped memory. It is strongest for assistants that must remember user facts, chat history, business data, relationships, and temporal context across sessions.

Skip it when you only need RAG storage. Qdrant, Weaviate, and Pinecone fit better when the problem is vector search over documents.

Best plan guidance: use the free 10k-credit route to test memory quality. Flex is the first self-serve production plan. Flex Plus is better when observations, custom extraction instructions, webhooks, analytics, and longer API logs matter. Enterprise is for SLA, BYOK, BYOC, HIPAA, audit logs, and custom rate limits.

Key Facts


Core job	Agent memory, temporal context graphs, and context assembly
Free path	10,000 credits/month for prototyping
Flex	$125/month or $1,250/year, 50,000 credits/month
Flex Plus	$375/month or $3,750/year, 200,000 credits/month
Enterprise	Custom credits, negotiated rates, SLA, BYOK, BYOC, HIPAA, audit logs
Credit unit	Episode size, starting at 1 credit per 350 bytes
Open-source related project	Graphiti is Apache-2.0 licensed

When To Pick Zep

Your agent needs memory beyond chat history. Zep is useful when facts, relationships, and context need to persist across sessions.
Temporal context matters. The platform is designed around context graphs that can track changing facts over time.
You need managed memory infrastructure. Zep Cloud avoids building memory extraction, graph storage, retrieval, and production operations from scratch.
You need enterprise deployment options. Enterprise adds SLA, guaranteed rate limits, audit/API logs, HIPAA BAA, BYOK, BYOC, support, and custom rates.
You are testing memory as a capability. The free 10k-credit path gives teams a lower-risk way to test before a production commitment.

When To Pick Something Else

Managed or open-source memory fork: Mem0 when the buyer wants a managed cloud plus Apache-2.0 OSS route around memories.
Vector search: Qdrant, Weaviate, or Pinecone when the job is retrieval over documents and embeddings.
Agent orchestration: LangGraph, Mastra, or CrewAI when the team needs the workflow/runtime layer rather than memory storage.
Observability and evals: LangSmith, Braintrust, or Langfuse when debugging and release-quality evidence are the priority.
Search and extraction: Tavily or Firecrawl when the missing data comes from live web context.

Pricing

Zep pricing was checked on June 28, 2026 against the official pricing page.

Plan	Price	Included shape	Buyer fit
Free prototyping	10,000 credits/month	Limited free usage for tests	First memory-quality checks
Flex	$125/month or $1,250/year	50,000 credits/month, 600 requests/minute, 5 projects, 1-day API logs	Early production agent memory
Flex Plus	$375/month or $3,750/year	200,000 credits/month, 1,000 requests/minute, 10 projects, observations, custom extraction, webhooks, analytics, 7-day API logs	Production memory with more controls
Enterprise	Custom	Custom credits, negotiated rates, SLA, unlimited projects, SOC 2 Type II, HIPAA BAA, 1-year audit/API logs, BYOK, BYOC	Regulated and high-volume teams

Credits are consumed by Episode size. Zep lists one credit for an Episode up to 350 bytes, another credit for each additional 350 bytes or part, and one eighth of a credit for webhook invocations where available. Memory, retrieval, storage, threads, users, and graph storage are listed as unmetered.

Failure Modes

Bad memory is worse than no memory. Incorrect, stale, or over-broad facts can make an agent feel confident for the wrong reason.
Consent and deletion matter. Agent memory can store personal facts, preferences, account context, and business data.
Credit use depends on Episode design. Payload size, ingestion volume, webhook usage, and extraction behavior drive cost.
A memory layer is not a full agent stack. You still need orchestration, tools, evals, logs, permissions, and fallback behavior.
Enterprise controls may be non-optional. BYOK, BYOC, HIPAA, audit logs, and retention terms can decide fit for regulated workloads.

Methodology

This page was produced by the aipedia.wiki editorial pipeline. Scoring follows the four-dimension rubric at /about/scoring/ (Utility x Value x Moat x Longevity, unweighted average). Last verified 2026-06-28 against Zep pricing, official site metadata, docs concepts, and Graphiti licensing.

FAQ

Is Zep free? Zep lists a free prototyping route with 10,000 credits/month. Production plans start with Flex and Flex Plus, while Enterprise is custom.

How does Zep charge? Zep uses credits based on Episode size. An Episode can be a chat message, JSON payload, or block of text.

Zep vs Mem0? Both target agent memory. Zep leans into production context graphs and commercial deployment options. Mem0 is a strong comparison when the buyer wants a managed platform plus an Apache-2.0 open-source route.

Sources

Zep official site: product positioning and context engineering claims
Zep pricing: Flex, Flex Plus, Enterprise, credits, limits, and deployment options
Zep docs concepts: product concepts and memory/context model
Graphiti repository: open-source context graph project
Graphiti license: Apache-2.0 license

Category: AI Infrastructure · AI Automation
Alternatives: Mem0 · Qdrant · Weaviate · LangGraph

Share LinkedIn

Was this review helpful?

Embed this score on your site Free. Links back.

HTML

<a href="https://aipedia.wiki/tools/zep/" target="_blank" rel="noopener"><img src="https://aipedia.wiki/badges/zep.svg" alt="Zep on aipedia.wiki" width="260" height="72" /></a>

Markdown

[![Zep on aipedia.wiki](https://aipedia.wiki/badges/zep.svg)](https://aipedia.wiki/tools/zep/)

Badge value auto-updates if the editorial score changes. Attribution via the link is required.

Cite this page For journalists, researchers, and bloggers

News writers

According to aipedia.wiki Editorial at aipedia.wiki (https://aipedia.wiki/tools/zep/)

APA

aipedia.wiki Editorial. (2026). Zep: Editorial Review. aipedia.wiki. Retrieved July 2, 2026, from https://aipedia.wiki/tools/zep/

MLA 9

aipedia.wiki Editorial. "Zep: Editorial Review." aipedia.wiki, 2026, https://aipedia.wiki/tools/zep/. Accessed July 2, 2026.

Chicago

aipedia.wiki Editorial. 2026. "Zep: Editorial Review." aipedia.wiki. https://aipedia.wiki/tools/zep/.

BibTeX

@misc{zep-editorial-review-2026,
  author = {{aipedia.wiki Editorial}},
  title = {Zep: Editorial Review},
  year = {2026},
  publisher = {aipedia.wiki},
  url = {https://aipedia.wiki/tools/zep/},
  note = {Accessed: 2026-07-02}
}

Spotted an error or want to share your experience with Zep?

Every tool page is re-verified on a recurring cycle, and corrections land faster when readers flag them directly. If you spot a stale fact, a missing capability, or have used Zep and want to share what worked or didn't, the editorial desk reviews every message sent through this form.

Email editorial@aipedia.wiki

Report outdated info Help us keep this page accurate