Mem0
Memory layer for AI agents and assistants, hosted or self-hosted
Mem0 is teaching coding agents to search memory less and trust it more.
◆Recent moves
- 1mo ago
Mem0 Node SDK (v3.3.1)
Stops OpenAI defaults leaking into non-OpenAI provider configs and fixes Turbopuffer filters. Cleanup after the 3.3.0 profiles release.
View source ↗ - 1mo ago
Mem0 Python SDK (v2.2.1)
add() now reports only memories actually stored and raises VectorStoreError when nothing lands, instead of returning phantom records. A correctness change that matters for anyone trusting memory writes.
View source ↗ - 1mo ago
Mem0 DeepSeek Plugin (v0.3.2)
The DeepSeek plugin stops pushing agents to search before every answer and reframes injected memories. Part of a cross-plugin shift to fewer, more targeted recalls.
View source ↗ - 1mo ago
Mem0 Pi Agent Plugin (v0.3.2)
The Pi agent plugin's context loader now makes one search instead of two to four parallel ones. The clearest cost reduction in the coordinated recall retune.
View source ↗ - 1mo ago
Mem0 OpenClaw Plugin (v1.2.1)
Tool description wording change for OpenClaw, with recall strategies unchanged. The same policy edit applied with minimal footprint.
View source ↗ - 1mo ago
Mem0 OpenCode Plugin (v0.4.1)
OpenCode drops system-context lines that mandated parallel searches before responding. Aligns the plugin with the new less-is-more recall policy.
View source ↗