Introducing Atlas-2-Small
AeroAI has released Atlas-2-Small, a new small language model built as part of our work on practical aviation-focused AI.
Today, AeroAI is introducing Atlas-2-Small, our newest small language model. The release marks the next step in our work to build practical AI systems for aviation and other specialized tasks.
Atlas-2-Small is made by AeroAI and is designed around the idea that useful AI does not always need to be enormous. A smaller model can be easier to develop, evaluate, deploy, and adapt for focused workflows.
This release also reflects a broader change in how we are building our models. Instead of treating a general-purpose model as the finished product, we are continuing to experiment with aviation-specific data, instruction formats, tool use, evaluation, and careful handling of uncertainty.
Atlas-2-Small is part of that ongoing development. Its release gives us a new foundation to evaluate, improve, and use in future AeroAI projects while we continue working on larger and more specialized systems.
As with any AI system used around aviation, Atlas-2-Small should be treated as an assistive technology rather than an authority. Aviation information can be time-sensitive, aircraft- and airport-specific, and safety-critical. Important answers should be checked against current authoritative sources and qualified human judgment.
We will continue evaluating Atlas-2-Small and improving our aviation AI stack as development progresses. This release is another step toward making specialized aviation AI more capable, practical, and useful.
