Home / Directory / Agents & orchestration / AI agent memory / Letta
Letta
San Francisco company spun out of UC Berkeley by the creators of MemGPT that builds open-source tools for AI agents with long-term memory that learn over time.
Developers building agents that must remember users and past work across sessions, who want open-source code and visible memory.
Teams that want a simple memory API bolted onto an existing agent, or buyers who need a large vendor with enterprise references.
Verdict
Letta is a San Francisco company, spun out of UC Berkeley, that builds AI agents with long-term memory developers can inspect and edit. It suits developers who want agents that remember and improve across sessions. Its MemGPT research helped make agent memory a standard design problem.
Score Breakdown
How Letta scores in the categories that matter to its buyers.
Pricing
Letta publishes free and Pro plans for individuals and usage-based developer plans for teams building on its API.
| Plan | Price | What stands out |
|---|---|---|
| Free | $0 / month | Up to 3 stateful agents, bring your own model keys |
| Pro | $20 / month | Up to 20 stateful agents, Letta Auto quota, pay-as-you-go overage |
| Developer | Usage-based | Unlimited agents through the Letta API |
| Teams / Enterprise | Custom | Shared agents and team permissions |
The Field at a Glance
How Letta compares with other AI agent memory vendors we reviewed, by Overall Score and relative typical engagement cost.
Letta scores 7.8 in AI agent memory, under Mem0 (8.0) and Zep (7.9), level with Supermemory (7.8), and ahead of Cognee (7.7). Relative cost lands lower for this subcategory.
Use-case matrix
| Use case | Fit | Notes |
|---|---|---|
| Long-term memory | Strong | Agents keep memory across sessions. |
| Visible, editable memory | Strong | Developers can inspect and change what an agent remembers. |
| Open source | Strong | Code is public on GitHub. |
| Free start | Strong | Free plan with bring-your-own model keys. |
| Simple drop-in memory API | Mixed | Built around full agents. |
| Enterprise track record | Weak | A young company founded in 2024. |
Who it’s for
Good fit
- Developers building personal assistants
- Teams building agents for long projects
- Researchers studying agent memory
Poor fit
- Teams wanting a one-line memory add-on
- Buyers needing large enterprise references
- Non-technical teams
Review Excerpts
Below are excerpts from public reviews. Our team scoured public reviews, forums, and chat rooms to get a balanced view of customers' experience with this company. Paid reviews and pay-for-play sites such as Clutch were excluded.
Letta lists Bilt, 11x, Kognitos, and Hunt Club among its featured case studies.
Letta's free plan lets developers start with their own model keys, including coding plans from ChatGPT and other providers, for up to 3 stateful agents.
Letta says heavy coding users typically spend about $100 a month or more in usage, and power users can pass $200 a month, beyond the $20 Pro plan.
Company Facts
- Origin
- Spun out of UC Berkeley's Sky Computing Lab, makers of MemGPT (PR Newswire)
- Headquarters
- San Francisco (letta.com)
- Seed funding
- $10M led by Felicis, September 2024 (PR Newswire)
- Founders
- Charles Packer (CEO) and Sarah Wooders (CTO) (PR Newswire)
- License
- Open source on GitHub (GitHub)
- Website
- letta.com
Key Leadership
Charles Packer
Co-founder and CEO
Methodology
This page is an independent evaluation of Letta for buyers comparing options in AI agent memory. AI Industry Reviews accepts no sponsorships, advertising, or pay-for-placement fees. Letta did not pay for this review.
What we scored
The headline number is an Overall Score on a 0-10 scale. Eight criteria sit under it in two groups.
- Buyer outcomes (for AI agent memory)
- Memory design
- Open-source access
- Cost to start
- Enterprise maturity
- Company & commercial
- Innovation & product leadership
- Project management & communication
- Pricing
- Contract fairness
Pricing measures whether the price looks fair for the value delivered, including packaging and renewal friction that show up in real buying cycles.
Score Composition
| Input | Weight | What it covers |
|---|---|---|
| Reviews | 40% | A proprietary read of what practitioners say about likes, complaints, and day-to-day use, including public review sites, forums, and private chat rooms. Paid reviews and pay-for-play sites such as Clutch are out of scope. |
| Product | 35% | Hands-on look at screens and workflows. |
| Pricing | 15% | Whether the price looks fair for what you get. |
| Docs & training | 10% | Docs, tutorials, and training material. |
How we balanced the evidence
The Overall Score is the simple average of the eight criteria. Recommendation language follows that score and the fit pattern described above.
Scope
Letta is graded here as AI agent memory. Criteria scores can move as more review volume and product checks are added.