Fermat's Last Token is designed for individuals and small teams that want to reduce Claude Code costs without changing how they work.
We do not store your prompts, code, tool outputs, or agent traces. We retain only content-free operational data, such as token counts per request and compression statistics. However, Fermat's context optimization and prose compression temporarily route selected context through our cloud-hosted compression models.
We recognize that some organizations cannot permit source code or agent context to be processed outside their existing infrastructure, even when that data is not retained.
Introducing Mersenne
Mersenne brings Fermat's optimized agent tools and tool-output compression to enterprise environments while performing all Quotient optimization locally. Your prompts, code, tool outputs, and agent traces are never sent to Quotient Labs. Only content-free authentication, billing, token-count, and compression statistics leave your machine.
Your requests still go to the model provider you have configured for Claude Code, such as Anthropic or Vertex AI. Mersenne does not introduce another cloud data processor into that path.
For organizations requiring a completely on-premises deployment, private infrastructure, or custom data-retention controls, we're more than happy to discuss further.
Talk to us
To discuss access, deployment requirements, or an on-premises installation, contact us at founders@quotientlabs.com.