Letta
Framework and cloud for stateful AI agents with persistent memory, plus a CLI coding agent.
How the 5.5 was reached
- Quality of output6/10
Independent benchmark shows about 83% on LoCoMo memory tests; only one independent source, plus vendor case studies. dev.to
- Ease of use6/10
One-line npm install is listed, but docs page was not found and no onboarding evidence. letta.com
- Pricing value8/10
Free plan with 3 agents, Pro $20/month, usage-based developer plans at $0.10 per agent plus LLM costs. docs.letta.com
- API and integration quality2/10
No API or integrations found on the homepage, pricing page or docs search; docs page not found. letta.com
- Solves the problem it claims to solven/a
no evidence found
5.5/10 is the mean of the 4 criteria that apply. Scored on the five-criteria rubric v1, 29 September 2026. How scoring works
Pricing
- 3 stateful agents
- Remote sandboxes
- Letta Auto quota
- Letta Auto with quotas
- $0.10 per active agent per month
- Pay-as-you-go LLM usage
Key Features
- Stateful agents with memory that persists across conversations
- CLI installed with npm i -g @letta-ai/letta-code
- Cloud agents with GitHub and Slack integrations
- Channels for Slack, Telegram, Discord, WhatsApp and Signal
- Self-hosting via an app server, and MCP support
Pros & Cons
Pros
- Free plan and a low $20 monthly Pro tier
- Usage-based developer pricing at $0.10 per agent
- Memory-focused design; one independent benchmark shows about 83% on LoCoMo
- One-line npm install
Cons
- Only one independent benchmark source for output quality
- API and integration evidence scored low in the review
- No clear onboarding evidence
- Pricing mixes seats, agents and LLM costs, which is harder to forecast
Suits developers experimenting with agents that need long-term memory, and the free plan makes that cheap to try. Teams that need firm integration docs or proven results should hold off.
Try Letta →Added to scored.tools on
Competitors to Letta
Other tools in the coding category worth comparing.
Claude-Mem
7.8/10AI memory archive that captures and organizes your coding assistant's decisions and context across sessions.
LangGraph
8.0/10Low-level orchestration framework for building, managing, and deploying stateful AI agents.
Acontext
6.4/10Skill memory layer for AI agents that learns from runs and creates reusable Markdown skill files.
LangChain
8.4/10Open source framework for building LLM-powered agents and applications with prebuilt architecture and integrations.