

Cognee
β Text source-checked Nov 13, 2026Open-source agent memory platform for LLM agents.
Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked How we rate
In short
Cognee is an open-source agent memory platform using graph and vector retrieval. It offers a free tier and supports local or cloud deployment for persistent LLM context.
Cognee is an open-source agent memory platform that builds persistent memory across sessions using graph, vector, and relational retrieval. Cognee connects scattered data sources into a single memory layer for agents. Cognee provides agents with the ability to retain experience and context across runs.
Cognee generates ontologies that agents follow to adhere to domain rules. Cognee offers first-party integrations for Claude Code, Cursor, LangGraph, and OpenClaw, along with an MCP server for compatible agents.
Cognee Cloud supports multi-tenant deployments with isolated workspaces and per-session memory for serving multiple users or customers. Cognee can run locally for development or be self-hosted for teams requiring infrastructure control and private deployments.
The Cognee SDK is open source on GitHub, with options for hosted Cognee Cloud or enterprise deployment. Cognee Cloud operates on the gpt-oss-120b open-weight model from OpenAI.
The vendor says that Cognee is GDPR-compliant, encrypts data at rest and in transit, and supports air-gapped and bring-your-own-cloud deployments.
According to Cognee: Cognee is an open-source agent memory platform that builds persistent memory across sessions using graph, Vector, and relational retrieval. Its pages list this strength: Cognee is a free open-source platform providing persistent long-term memory for AI agents.
β The AI Tool Bible editorial team
Pros
- β Cognee is a free open-source platform providing persistent long-term memory for AI agents.
- β Cognee provides first-party integrations for major agent frameworks and an MCP server.
- β Cognee Cloud supports multi-tenancy with isolated workspaces and per-session memory.
- β The vendor says that Cognee supports GDPR compliance, encryption, and air-gapped or BYOC deployments.
Facts
- Platforms
- linux, macos, windows
- API available
- Yes
- Free trial
- Yes (not source-checked)
- Open source
- Yes (license: Apache-2.0)
- GitHub stars
- 33,287 (checked Nov 13, 2026) (not source-checked)
- Last GitHub push
- Nov 13, 2026 (not source-checked)
- First commit
- Aug 2023 (not source-checked)
- Model used
- Multi-model (Claude, OpenAI, others) (not source-checked)
- Vendor pages
- Changelog
- Package downloads (30 d)
- 86,048 Β· pypi cognee Β· checked Nov 9, 2026 (not source-checked)
- Hacker News
- 3 stories Β· top 3 points Β· last Sep 26, 2026 Β· checked Nov 9, 2026
- Earliest Wayback capture
- 2024-03-24 (earliest archived capture of the site, not a launch date)
Where this tool sits
Use cases
Recent updates from Cognee
- Meet the new cognee: smarter memory, lower costs, and control for your team
- cognee Joins UC Berkeley Xcelerator's 2026 Agentic AI Cohort
- cognee 1.0: The Open-Source Memory Platform for AI Agents
- Claude Code's Leak Reveals Anthropic's Obsession with Cognee
- Cognee Raises $7.5M Seed to Build Memory for AI Agents
Frequently asked
- How much does Cognee cost?
- Cognee offers a freemium model. The free plan costs $0 per month and includes one workspace with 1 million tokens. Paid options include hosted Cognee Cloud or enterprise deployment, though specific pricing for these tiers is not documented in the provided text.
- Which AI models does Cognee support?
- Cognee is multi-model, supporting Claude, OpenAI, and others. Cognee Cloud specifically operates on the gpt-oss-120b open-weight model from OpenAI. You can also bring your own cloud or run it locally for development.
- What integrations are available for Cognee?
- Cognee provides first-party integrations for Claude Code, Cursor, LangGraph, and OpenClaw. It also offers an MCP server for compatible agents. The SDK is open source on GitHub, allowing for flexible integration into existing agent workflows.
- Can I self-host Cognee for privacy?
- Yes, Cognee can run locally for development or be self-hosted for teams requiring infrastructure control. It supports air-gapped and bring-your-own-cloud deployments. The vendor states it is GDPR-compliant and encrypts data at rest and in transit.
- What are the main use cases for Cognee?
- Key use cases include building a Company Brain, providing agent memory, creating GraphRAG pipelines, and powering customer support agents. It connects scattered data sources into a single memory layer, allowing agents to retain experience and context across runs.