# Mem0

> Mem0 builds persistent memory infrastructure for AI applications and agents. Its open-source engine and managed platform extract useful facts from interactions, update them as circumstances change, and retrieve relevant context across sessions. Developers use it to personalize assistants without repeatedly sending entire conversation histories; OpenMemory extends the portability idea to compatible AI tools.

- **Founded:** 2023
- **Headquarters:** San Francisco, United States
- **Founders:** Taranjeet Singh (Co-founder and CEO), Deshraj Yadav (Co-founder and CTO)
- **Products:** Mem0 Open Source, Mem0 Platform, OpenMemory MCP, OpenMemory Chrome Extension, DolphinBench
- **Notable:** Y Combinator Summer 2024 participant., Announced approximately $24 million in combined seed and Series A funding on October 28, 2025., Reported 41,000 GitHub stars and 14 million Python package downloads at the October 2025 announcement.

## Products & services

- **Mem0 Open Source** — Apache-2.0 memory engine available as a library or self-hosted server; developers configure models, embeddings and storage. Self-hosting still incurs infrastructure and model costs.
- **Mem0 Platform** — Managed memory API and SDKs for extracting, storing, updating and retrieving context, with governance and enterprise options.
- **OpenMemory MCP** — Launched as a local-first memory server and UI for sharing context among compatible MCP clients. Local storage does not by itself establish that every configured model call stays on-device.
- **OpenMemory Chrome Extension** — Browser extension announced for carrying user-controlled context between supported browser AI assistants; includes memory inspection and editing.
- **DolphinBench** — Open agent-memory benchmark released September 22, 2026, evaluating actions in simulated applications alongside cost and latency; a research resource rather than a subscription tier.

## Achievements

- Y Combinator Summer 2024 participant.
- Announced approximately $24 million in combined seed and Series A funding on October 28, 2025.
- Reported 41,000 GitHub stars and 14 million Python package downloads at the October 2025 announcement.
- Reported API calls rising from 35 million in Q1 to 186 million in Q3 2025.
- April 2025 research reported 26% relative improvement over OpenAI’s memory baseline on LoCoMo, with 91% lower p95 latency and over 90% token-cost savings against full-context processing; benchmark-specific results.
- OpenNote case study reported a 40% reduction in prompt token usage and a two-day integration; vendor-published customer results.

## Latest updates

- **2026-09** — Released DolphinBench on September 22 to measure memory-dependent agent actions, cost and latency across simulated workflows.
- **2026-09** — Announced Vercel Marketplace availability on September 16, including scoped project setup and billing through Vercel.

## Links

- Website: https://mem0.ai/
- LinkedIn: https://www.linkedin.com/company/mem0/
- Twitter/X: https://x.com/mem0ai
- GitHub: https://github.com/mem0ai
- YouTube: https://www.youtube.com/@mem0dotai

---

Profile page: https://yespress.io/mem0
Published by YesPress — https://yespress.io
Last updated: 2026-10-01