Letta
Stateful agents with memory you own and port across models
What is Letta?
Letta builds stateful agents with persistent, portable memory; their flagship Letta Code is a memory-first coding agent (CLI, desktop, SDK) where the agent's accumulated knowledge moves with you across model providers. Sold to developers and AI researchers who want agents that learn over time rather than stateless chatbots. Spun out of UC Berkeley's MemGPT research.
Tools for building, hosting, testing, observing, connecting, and giving memory or computer access to AI agents.
See the full Agent Infrastructure guide to compare more tools, buyer criteria, and related workflows.
Use cases to evaluate
Long-running coding assistant that retains project context across sessions
Personal AI that learns your preferences and ports between Claude/GPT/Gemini
Research on continual learning and inference-time compute
Building distributed agents that share memory across devices
Fit to evaluate
Developers wanting a memory-first alternative to Claude Code/Cursor
AI researchers experimenting with stateful agents
Power users who want to switch LLMs without losing their assistant
Teams building agents where long-term memory is the product
Business fit
Right for you if you want an agent that remembers prior sessions, can switch underlying LLMs without losing context, and you are comfortable in terminal/SDK workflows. BYO-key support means you can avoid Letta's credit markup. Skip if you only need a single-turn or single-session agent or if memory portability is not worth managing a separate stateful service.
How to evaluate Letta
Use this category when a business wants agents that do work across tools, APIs, browsers, and data sources.
Confirm the exact workflow
Map Letta to one concrete workflow first, such as long-running coding assistant that retains project context across sessions. Avoid buying before the owner, trigger, output, and success metric are clear.
Check category fit
Compare tool-calling, memory, browser automation, evals, observability, and deployment controls.
Compare practical alternatives
Shortlist Letta against Orgo, Browser Use, Browserbase so the decision is based on fit, effort, and workflow ownership rather than brand recognition alone.
Validate cost and rollout effort
Free: limited agents and limited Letta Auto usage, BYO API keys supported. Pro $20/mo: weekly+monthly Letta Auto quota, up to 20 stateful agents, PAYG overages at passthrough (no markup) provider pricing. Docs note casual users ~$100+/mo, power users ~$200+/mo. Separate API Plan for automated/external use (pricing not published in this doc). Also confirm implementation time, support needs, and whether the technical setup matches your team.
Compare Letta with alternatives
Use this quick comparison before booking demos or moving data into a new system.
| Primary workflow | Long-running coding assistant that retains project context across sessions, Personal AI that learns your preferences and ports between Claude/GPT/Gemini |
|---|---|
| Best-fit team | Developers wanting a memory-first alternative to Claude Code/Cursor, AI researchers experimenting with stateful agents |
| Implementation effort | Technical setup and maintenance profile |
| Pricing check | Free plan + paid plans |
| Closest alternatives | OrgoBrowser UseBrowserbaseHyperbrowser |
Letta pricing
| Model | Free plan + paid plans |
|---|---|
| Snapshot | Free: limited agents and limited Letta Auto usage, BYO API keys supported. Pro $20/mo: weekly+monthly Letta Auto quota, up to 20 stateful agents, PAYG overages at passthrough (no markup) provider pricing. Docs note casual users ~$100+/mo, power users ~$200+/mo. Separate API Plan for automated/external use (pricing not published in this doc). |
| Checked |
Common questions about Letta
What is Letta?
Letta builds stateful agents with persistent, portable memory; their flagship Letta Code is a memory-first coding agent (CLI, desktop, SDK) where the agent's accumulated knowledge moves with you across model providers. Sold to developers and AI researchers who want agents that learn over time rather than stateless chatbots. Spun out of UC Berkeley's MemGPT research.
What is Letta used for?
Common use cases: Long-running coding assistant that retains project context across sessions; Personal AI that learns your preferences and ports between Claude/GPT/Gemini; Research on continual learning and inference-time compute; Building distributed agents that share memory across devices.
How much does Letta cost?
Free: limited agents and limited Letta Auto usage, BYO API keys supported. Pro $20/mo: weekly+monthly Letta Auto quota, up to 20 stateful agents, PAYG overages at passthrough (no markup) provider pricing. Docs note casual users ~$100+/mo, power users ~$200+/mo. Separate API Plan for automated/external use (pricing not published in this doc).
Who is Letta best for?
Letta fits Developers wanting a memory-first alternative to Claude Code/Cursor, AI researchers experimenting with stateful agents, Power users who want to switch LLMs without losing their assistant, Teams building agents where long-term memory is the product. Right for you if you want an agent that remembers prior sessions, can switch underlying LLMs without losing context, and you are comfortable in terminal/SDK workflows. BYO-key support means you can avoid Letta's credit markup. Skip if you only need a single-turn or single-session agent or if memory portability is not worth managing a separate stateful service.
What are alternatives to Letta?
Common alternatives to Letta include Orgo, Browser Use, Browserbase, Hyperbrowser, Steel, Anchor Browser.