Research

Don't just study what AI can do. Study how it learns.

Our open-source playground for experimenting with how AI learns.

21 languages · 330 models · thousands of training checkpoints

Beetle brings together open model weights, training checkpoints, datasets, tokenisers, a training framework and analysis tools, giving you the resources to go beyond evaluating finished models and start investigating how they develop.

Stop treating AI models as black boxes. Start exploring how they develop, and help us build a better scientific understanding of language models for safer, fairer AI.

For everyone

What can YOU do with Beetle?

Whether you are interested in building small language models, AI interpretability, building better AI systems for more languages, or researching how humans learn language, the Beetle model suite and our open-source training, data and analysis tools let you ask cool questions.

Beetle, built for AI researchers, multilingual NLP and model builders.

For AI enthusiasts, researchers & model builders

What you can do with Beetle
  • AI enthusiasts. Get hands-on with open models and explore questions that go beyond benchmark scores.
  • AI researchers. Investigate model internals: tokenisation, memorisation, representations and interpretability. Learn when capabilities emerge during training and test hypotheses about learning dynamics, exploring what changes as models see more data (100M, 2B and 24B training tokens).
  • Multilingual NLP researchers. A suite of monolingual and bilingual models with checkpoints, from Finnish and Basque to Hindi, Japanese and Korean, for studying how multilingual transfer happens.
  • Model builders. Train your own small language models, experiment with different data mixtures and language-exposure curricula, and see what happens when you change how a model learns.
Beetle, built for cognitive scientists, applied AI researchers and anyone curious about AI.

Beetle isn't just for AI engineers

What you can do with Beetle
  • Compare AI with human learning. Compare the behaviour of language models with human cognitive data.
  • Learn the science behind AI models. Use Beetle for hands-on exploration of the science behind language models: analyse our models and learn how to train your own.

With 330 monolingual and bilingual models, spanning 21 first languages, five language-exposure curricula, and approximately 30 checkpoints per model, Beetle gives researchers an unprecedentedly controlled way to investigate how models acquire and process multiple languages. Our open-source Beetle Framework brings together open model weights, training checkpoints, data, tokenisers and analysis tools in one research suite.