Your Letta agents use large language models (LLMs) to reason and take actions. These model requests are what we count toward your monthly requests quota. Standard models (GPT-4o mini, Gemini 2.5 Flash, etc.) are faster and more economical. They’re ideal for simple tool calling and basic chat interactions. Premium models (GPT-4.1, Claude Sonnet 4, etc.) offer enhanced capabilities for complex agentic tasks. They excel at multi-step tool sequences and tasks requiring advanced reasoning.

Each agent “step” or “action” counts as one model request. Complex tasks (such as deep research) may require multiple requests to complete. You can control request usage via tool rules that force the agent to stop on certain conditions.

Your Letta agents use large language models (LLMs) to reason and take actions. These model requests are what we count toward your monthly requests quota. Request quotas refresh every month.

Yes, you can connect your own API keys for providers such as OpenAI, Anthropic, and Google Gemini. Model requests do not count towards your request quota if you bring your own LLM API key and select your custom provider in the ADE model dropdown.

Letta Cloud is our fully-managed service for stateful agents. While Letta can be self-hosted, Letta Cloud eliminates all infrastructure management, server optimization, and system administration so you can focus entirely on building agents. Letta Cloud also includes advanced features for power users, such as agent templates, monitoring, and observability.

Yes, Letta Cloud supports agent file, which allows you to move your agents freely between self-hosted instances of the Letta open source and Letta Cloud.

Yes, we offer discounts for qualifying startups. Submit a request using this form.

We have an active developer community on Discord, join to chat with the Letta developer team and other Letta users!