Zep
Generative AI infrastructure software
Generative AI software
- Features
- Ease of use
- Ease of management
- Quality of support
- Affordability
- Market presence
Take the quiz to check if Zep and its alternatives fit your requirements.
Pay-as-you-go
Free Trial unavailable
Free version
Small
Medium
Large
- Information technology and software
- Professional services (engineering, legal, consulting, etc.)
- Healthcare and life sciences
What is Zep
Zep is an open-source memory layer and retrieval service for LLM applications that stores, summarizes, and retrieves conversation and user context. It is used by developers building chatbots and AI assistants who need long-term memory, session management, and context injection into prompts. Zep provides APIs/SDKs and a server component to persist chat histories, generate summaries, and support retrieval workflows that can integrate with vector databases and application backends.
Purpose-built LLM memory layer
Zep focuses on conversation memory primitives such as chat history storage, summarization, and retrieval for context windows. This specialization can reduce the amount of custom code required compared with building memory management directly into an application. It fits common assistant patterns such as per-user memory, per-session context, and message-level metadata.
Open-source and self-hostable
Zep is available as open source, which supports code inspection and local deployment. Self-hosting can help teams meet data residency or internal security requirements by keeping conversation data in their own environment. It also enables customization of storage backends and operational controls to match internal standards.
API-first integration approach
Zep exposes APIs/SDKs designed to be called from application services, making it straightforward to integrate into existing LLM orchestration stacks. It can sit alongside retrieval and search components by providing a dedicated store for conversational context. This separation of concerns can simplify architecture when multiple applications share the same memory service.
Narrow scope beyond memory
Zep primarily addresses memory and conversational context, not end-to-end agent building, analytics, or full data science workflows. Teams may still need separate components for document ingestion, evaluation, observability, and governance. This can increase the number of tools to operate in production.
Operational overhead for self-hosting
Running Zep as a service introduces deployment, scaling, monitoring, and backup responsibilities. Production use typically requires planning for persistence, latency, and high availability, especially when memory is shared across many users. Organizations without platform engineering support may prefer managed alternatives where available.
Enterprise controls may be limited
Compared with broader enterprise AI platforms, Zep may offer fewer built-in features for centralized governance, role-based administration, audit logging, and compliance reporting. Some controls may need to be implemented through surrounding infrastructure (API gateways, IAM, logging pipelines). This can lengthen time-to-compliance for regulated deployments.
Plan & Pricing
| Plan | Price | Key features & notes |
|---|---|---|
| Flex | $25 per month | 20,000 credits included; auto-topup at 20% (+20,000); then $25 / 20,000 credits; 600 requests per minute; 5 Projects; 10 custom entity & edge types; Unlimited memories, retrieval & users. |
| Flex Plus | $475 per month | 300,000 credits included; auto-topup at 20% (+100,000); then $125 / 100,000 credits; 1,000 requests per minute; 5 Projects; 20 custom entity & edge types; Custom extraction instructions; Webhooks; API logs (7 days); Unlimited memories, retrieval & users. |
| Enterprise | Custom pricing | Contact sales; includes SOC 2 Type II & HIPAA BAA options, custom limits & guaranteed rate limits, Slack support & dedicated account manager, API & audit logs, SLA guarantees; deployment options: Managed, BYOK, BYOM, BYOC. |
Usage-based (Metered) pricing (vendor also offers a pay-as-you-go metered option):
Pricing model: Pay-as-you-go (Metered) Free tier: 2,500 messages free per month; 2.5 MB Graph Data free per month (limits apply). Rates: $1.25 per 1,000 messages (after free tier); $2.50 per MB of Graph Data (after free tier). Includes: Up to 5 Projects; in-app chat support with upgrade available.
(Notes: Zep’s site shows both credit-based Flex plans and a Metered pay-as-you-go option; Enterprise plans require contacting sales.)
Seller details
Zep AI, Inc.
San Francisco, California, United States
2023
Private
https://www.getzep.com/
https://x.com/zep_ai
https://www.linkedin.com/company/zep-ai
