Skip to content
NewMinniCode is out now, in Experimental Beta.Get MinniCode→
Minni AI
DemoEvidenceService
Products
Products
  • MinniMemoryMCPFlagship, early accessMemory for your AI agentSplits your agent’s memory into a small always-on part and files that load only when a task needs them.
  • MinniCodeExperimental BetaOpen-LLM optimized, low token coding agentA low token terminal coding agent, optimized for open LLMs. Reads, edits, tests and opens pull requests.
  • MinniTunerExperimental Beta, WindowsHardware optimization agentTunes how your PC uses its CPU, GPU and RAM. One plan you approve, every change undoable.
DocsMinniMemoryMCP docsInstall it, the commands, undoMinniCode docsInstall it, pick a model, the commandsMinniTuner docsThe tiers, Ask Minni, what is sent
See it workDemoThe full tour in one videoHow it worksEach product, step by step
Resources
LearnGuidesFAQnpm package
CompanyContactSupportPrivacyTermsData handlingSecurity

Guides

MinniMemoryMCP vs Mem0, Zep, supermemory and Letta

Mem0, Zep, supermemory and Letta solve one problem: an agent that forgets. They extract memories from conversations with a model, keep them in a hosted store, and retrieve them per query. MinniMemoryMCP solves a different one: an agent whose memory files have grown until every turn pays for them. It sorts the files you already have into a small always-on part and files that load on demand, on your computer, with no model and no account. You can run it beside any of the four.

Updated 2026-11-16

Two different problems#

An agent can have too little memory or too much. Too little: it forgets the user between sessions, so a memory service stores what it learns and hands it back when a query matches. Too much: its instruction and memory files (CLAUDE.md, AGENTS.md, the auto memory index) have grown, and all of it is sent with every turn whether the task needs it or not.

The four tools below are built for the first problem. MinniMemoryMCP is built for the second. The comparison is not which is better; it is which problem you have. Many Claude Code users have the second and are shopping for the first.

Side by side#

What it is, in its own wordsWhere the memory livesNeeds a model to build memoryFree tierClaude Code
MinniMemoryMCPAn MCP server that shrinks what Claude Code loads every turnYour own files, on your computer: the always-on part and .claude/OnDemandMemory/No. No model, no account, no networkFree during early accessYes: MCP server, four read-only commands, the Claude Code plugin
Mem0"The memory layer for AI agents and applications"Mem0’s hosted platform (API key from app.mem0.ai); on-prem on the Custom planYes: the free tier counts model calls, then uses your own LLM keyFree: 10,000 adds and 1,000 retrievals a month for 30 days, then 50,000 adds and 10,000 retrievals a month with your keyThrough its MCP integration (mem0-mcp); Claude Code is not named on its site
Zep"The enterprise context layer for reliable agents"Zep Cloud, Cloud with your own keys, or Bring Your Own Cloud in your VPCNot stated on its siteFree: 10,000 credits a month, 2 projects, 1 Context MCP Server seat; Flex $125 a month, Flex Plus $375 a monthListed, with Codex, Cursor, Copilot and Gemini CLI, through MCP or the SDKs
supermemory"The default engine for memory and continual learning for agents"Its hosted API and vector-graph database; self-hosting on Scale and EnterpriseYes: its own model, learner-1, extracts and consolidatesFree: $5 of credits a month; Pro $19, Max $100, Scale $399 a monthYes: a Claude Code plugin and an MCP server
Letta"The platform for building stateful agents", from the creators of MemGPTLetta Cloud, or your own Letta App Server; the agent harness is open sourceThe agent itself is the memory manager: it is an agent platform, not an add-onSee its pricing pageIts own harness and a hosted MCP page; Letta Code is its own coding agent

Sources: mem0.ai and its pricing page, getzep.com and its pricing page, supermemory.ai and its pricing page, docs.letta.com, all read on 7 November 2026. Prices change; their pages are the record.

What the four add#

  • Mem0 extracts memories from each interaction, condenses chat history into compact memories, and returns the ones that match a query, per user. It is a drop-in layer for an app you build.
  • Zep builds a temporal context graph from conversations and business data (Slack, Gmail, Notion, Snowflake and more), with policies on what each user and agent may read. It is aimed at companies.
  • supermemory runs its own model to extract and consolidate context per user, task or tenant, stores it in a vector-graph database, and injects matching memories into the model’s context at query time. Plugins for Claude Code and Cursor.
  • Letta is an agent runtime where the agent manages its own memory and keeps learning across sessions; MemGPT was the research it grew from. You build the agent on Letta rather than adding Letta to an agent.

All four put more into the context: the memories they retrieve are added to the turn. That is the point of them, and it is a cost you take on because the alternative is an agent that does not remember.

What MinniMemoryMCP does instead#

MinniMemoryMCP takes less out of the context. It reads the memory files Claude Code already loads, finds the sections that only some tasks need (histories, runbooks, past decisions, dated notes), and moves them into OnDemandMemory files with one line each in the always-on part saying when to open them. Nothing is reworded; every line lands somewhere; undo puts it all back.

  • It runs on your computer: no model inside it, no account, no key, no network call.
  • Over MCP it never writes: apply and undo return the command and the list of files, and Claude runs it through your own approval prompt.
  • A prompt hook places the one matching section into a turn when it clearly helps, so the agent answers in one call instead of opening files.
install it for Claude Code
$ npm i -g minnimemory
$ minnimemory install

With MinniMemoryMCP 1.6.0 on our test memory, the agent went looking in its memory 17 times instead of 89 over the same 87 answers, and the run used 47.2% fewer tokens.

We asked the same 30 questions in Claude Code 2.1.325 on Claude Sonnet 5.5, three full sessions a side: once with a test memory in the layout Claude Code itself writes, as it was, and once after MinniMemoryMCP 1.6.0 sorted it, with its server and hook on. Measured 15 November 2026: at API rates, a Claude Code session on our test memory cost $0.47 before MinniMemoryMCP 1.6.0 and $0.29 after on average, 37.7% less (each of the three sessions 33.9% to 40.0% less). A blind judge found 80 of 87 answers correct before and 81 after. The method is on the evidence page.

Which one you need#

IfThen
You build an app or agent that must remember its users across sessionsA memory service: Mem0 or supermemory for a layer you call, Zep for company data with access rules, Letta if the agent itself should own its memory
You use Claude Code and its memory files have grownMinniMemoryMCP: run init, read the plan, apply
BothBoth. A memory service feeds the agent; MinniMemoryMCP keeps what the agent loads every turn small. They touch different files
Your memory file is short and you edit it by handNeither yet. See CLAUDE.md vs MinniMemoryMCP for when a split starts to pay

What Claude Code carries between sessions on its own, with no add-on, is in How Claude Code remembers across sessions.

Try MinniMemoryMCP

Free, runs on your computer, and every change can be undone. Two commands install it for Claude Code; the docs cover Codex and the rest.

install it for Claude Code
$ npm i -g minnimemory
$ minnimemory install

More guides

  • How to reduce Claude Code token usage

    Where the tokens go each turn, and eight ways to spend fewer.

    Updated 2026-11-16 · Minni AI

  • Claude Code usage limit reached: what to do, and how to hit it less

    What each limit message means, what to do now, and how to send less every turn.

    Updated 2026-11-16 · Minni AI

  • CLAUDE.md too long: what Claude Code does with it, and how to split it

    What a long CLAUDE.md costs, the real size limits, and four ways to cut it.

    Updated 2026-11-16 · Minni AI

  • Claude Code context window keeps filling up: why it compacts, and what to do

    What fills the window, what compaction keeps, and how to compact less often.

    Updated 2026-11-16 · Minni AI

  • How to reduce Codex token usage

    Where Codex spends tokens, and seven ways to spend fewer.

    Updated 2026-11-16 · Minni AI

  • Codex usage limits: why they drain and how to use fewer tokens

    How the windows work, how to check what is left, and what uses the allowance up.

    Updated 2026-11-11 · Minni AI

  • AGENTS.md too large: what Codex cuts off and how to split it

    The 32 KiB cap, how to see which files load, and five ways back under it.

    Updated 2026-11-11 · Minni AI

  • How to lower a Claude Code or Codex API bill

    What a coding agent bills on an API key, and the levers that move it.

    Updated 2026-11-16 · Minni AI

  • Why Claude Code sessions get expensive

    Every turn re-sends the whole session. What makes that grow.

    Updated 2026-11-05 · Minni AI

  • Claude Code memory: CLAUDE.md vs MinniMemoryMCP

    What belongs in CLAUDE.md, and splitting the rest out by hand or with a tool.

    Updated 2026-11-16 · Minni AI

  • How Claude Code remembers across sessions

    What carries over between sessions, what does not, and how to keep it small.

    Updated 2026-11-16 · Minni AI

  • Where Claude Code keeps its memory

    Every memory path on your machine, what each file is for, and which ones load every turn.

    Updated 2026-11-07 · Minni AI

  • How to run a coding agent on free, open models

    What a free key gets you, what a local model can do, and where the tokens go.

    Updated 2026-10-25 · Minni AI

  • Get a free NVIDIA API key for a coding agent

    Where the key comes from, where to keep it, and how to run a coding agent on it.

    Updated 2026-11-07 · Minni AI

  • What a Windows PC optimization agent should and should not touch

    The settings worth changing, the ones that are placebo, and the ones that break things.

    Updated 2026-10-25 · Minni AI

GuidesMinniMemoryMCPHome
© 2026 Minni AI. All rights reserved.