What is Inserloft?

Inserloft is a company focused on building infrastructure for artificial intelligence.

Its goal is not only to create large language models, but to build a complete ecosystem around them, including:

• Frontier Models — the largest, most capable models Inserloft trains (the Kyro family).
• Baby LLMs — small, efficient, specialized models designed to run locally and cheaply.
• AI Platform — Pixel, a workspace for building, sharing, and deploying AI work.
• Open Source — tools, libraries, and open-weight models published by OpenGit.
• Developer Tools — SDKs, a CLI, and documentation for integrating Inserloft into real products.
• APIs — programmatic access to chat, embeddings, models, and files.
• AI Agents — Cleo and the automation built on top of it.
• Research — the lab behind architectures, training, alignment, and inference work.

Each of these pieces is designed to work together. A model trained by Research becomes part of the Models catalog, is hosted and versioned on the Pixel Platform, is exposed through the APIs, and powers products like Cleo. No layer is meant to stand alone.