gbrain vs MemPalace
Two of the top projects in Memory, side by side: who leads on each score part and who each one suits, then setup, license, activity and what each review found.
gbrain
Long-term memory store for coding agents, with sources and corrections
MemPalace
Local long-term memory store for agents with verbatim storage and semantic search
Verdict
gbrain ranks higher, 81 to 77. Of the 5 score parts we measure for both, gbrain leads on 3, MemPalace on 1 and 1 is even.
- AdoptionMemPalace leads by 43 points, 84 to 41: a wide gap.
- FreshnessEven at 100.
- Maintenancegbrain leads by 15 points, 98 to 83: a wide gap.
- Easy to rungbrain leads by 33 points, 83 to 50: a wide gap.
- Agent-readygbrain leads by 15 points, 85 to 70: a wide gap.
Pick gbrain if you are
Developers giving Claude Code, Codex or other agents persistent shared memory
- Keyless start: PGLite brain with keyword search, no server or Docker needed
- One brain shared across agents via stdio MCP or HTTP MCP with scoped OAuth tokens
Pick MemPalace if you are
Users of Claude Code, Codex or MCP agents wanting local persistent memory
- Raw retrieval scores 96.6% R@5 on LongMemEval with no LLM or API key
- Pluggable backends: Chroma, SQLite exact, Rust, Milvus, Qdrant, pgvector
| What we compare | gbrain | MemPalace |
|---|---|---|
| Score parts, out of 100 | ||
| Adoption | 41, known | 84, widely used |
| Freshness | 100, active | 100, active |
| Maintenance | 98, healthy | 83, healthy |
| Easy to run | 83, very easy | 50, easy |
| Agent-ready | 85, ready | 70, partly |
| Facts from GitHub and the README | ||
| Stars | 30.9k | 59.6k |
| License | MIT (permissive) | MIT (permissive) |
| Last commit | Oct 2026 | Oct 2026 |
| Last release | Oct 2026 | Oct 2026 |
| Language | TypeScript | Python |
| Docker | Yes | Yes |
| GPU | Not needed | Optional |
| arm64 or Apple Silicon | Mentioned | Mentioned |
gbrain
GBrain keeps explicit facts and their sources in a local PGLite or Postgres brain and exposes them to agents such as Claude Code, Codex, Cursor and Hermes over stdio or HTTP MCP. It starts keyless with keyword search; semantic search, cited synthesis, a typed knowledge graph and an overnight enrichment cycle are optional. It is installed with Bun from GitHub, not npm.
Who it is for: Developers giving Claude Code, Codex or other agents persistent shared memory
Strengths
- Keyless start: PGLite brain with keyword search, no server or Docker needed
- One brain shared across agents via stdio MCP or HTTP MCP with scoped OAuth tokens
- One-command setup for Claude Code and Codex, with dry-run and remove options
- Optional cited synthesis that notes gaps, plus a typed graph for relationship queries
Weaknesses
- Not on npm; the unrelated npm package named gbrain can shadow the real binary
- Always-on OpenClaw/Hermes setup needs a server with 8GB+ RAM plus paid API usage
- Local files and shared database credentials are not isolated by sources
- Remote put_page does not extract graph edges inline; HTTP needs explicit maintenance
- RAM ≥ 8 GB
- no GPU
- Compose
- Needs Bun 1.4.0+, PGLite or Postgres, Tailscale (optional, for remote MCP)
- Models: Optional cloud embedding, reranking, extraction and synthesis providers (unspecified)
MemPalace
MemPalace stores conversation history as verbatim text and retrieves it with semantic search, organized into wings (people, projects), rooms (topics) and drawers (original content). It ships a CLI, 47 MCP tools, a SQLite temporal knowledge graph, and auto-save hooks for Claude Code, Codex CLI and Cursor. ChromaDB is the default backend; sqlite_exact, rust_exact, Milvus, Qdrant and pgvector are selectable.
Who it is for: Users of Claude Code, Codex or MCP agents wanting local persistent memory
Strengths
- Raw retrieval scores 96.6% R@5 on LongMemEval with no LLM or API key
- Pluggable backends: Chroma, SQLite exact, Rust, Milvus, Qdrant, pgvector
- Benchmark results and reproduction commands are committed in the repository
- Optional OpenAI-compatible embedding endpoint, including local LM Studio, llama.cpp or Ollama
Weaknesses
- Native Termux install unsupported; Android needs a Debian PRoot container
- Stores verbatim text only; no summarization or fact extraction
- First embedding call downloads a model of about 80 to 300 MB
- GPU Docker image is not published and is x86_64-only
- GPU optional
- Docker + Compose
- Needs ChromaDB (default backend), ONNX Runtime, Optional: Milvus, Qdrant, PostgreSQL with pgvector
- Models: all-MiniLM-L6-v2, embeddinggemma-300m, OpenAI-compatible /v1/embeddings endpoints
More in Memory
Building with these?
We help teams pick an open-source AI stack and run it in production. Client work never changes a rank.