Models
Baby LLMs
Small, efficient models designed to run locally.
Baby LLMs are Inserloft's category of small language models, built with a very different goal from the Kyro frontier family.
A Baby LLM is designed to:
- Run locally, without requiring a hosted API
- Consume very little compute and memory
- Respond quickly, even on modest hardware
- Cost very little to run at scale
- Specialize in a narrow set of tasks rather than generalize broadly
Baby LLMs are not trying to compete with frontier-scale models on raw benchmark performance. They are meant to be efficient tools: fast enough and cheap enough to embed directly into products, developer tooling, and edge deployments.
Verty is the flagship example of a Baby LLM, and the first one released as open weights through OpenGit.