Evaluations
Evals in the real world
Evaluate models on real-world tasks and workflows to understand how they perform, where they fail, and whether they can deliver reliable results.
The RSI infrastructure behind frontier labs
Baking the next
discovery
Our work
From insight
to improvement.
We connect research, evaluation, and infrastructure to help frontier teams turn ideas into measurable progress.
Explore our products
Products
Discuss evaluations
Environments
Tasks and verifiers
Environments where models and agents can be tested, learn from feedback, and make measurable progress.
Infrastructure
Sandbox cluster
Run your autonomous research workloads in isolated sandboxes built for controlled experiments, with unified execution and traceable results.
Research
Ideas into progress.
and moreRSI Arena: Towards Recursive Self-Improvement with Human Feedback
A step toward recursive self-improvement with human feedback (RSIHF): ten AI agents develop models, people judge their outputs, and those judgments inform the next round of development, alongside capability and safety evaluations.
Reward Hacking Challenges Oversight of Autonomous Research Agents
Research agents can obtain high scores without making real improvements. We examine when AI reviewers catch these shortcuts and whether agents can use review feedback to conceal them.
Épi: Turning compute into verifiable improvement
Models provide capability. Environments make progress verifiable. Épi turns compute into improvement.
Models inside the R&D loop
Can models turn ideas into experiments, learn from the results, and make meaningful progress in research and engineering?
Can frontier models judge beauty?
An expert-curated benchmark for visual aesthetic judgment across fine art, photography, and illustration.
About Bake AI
Intelligence,
built to improve.
We believe there is always more to understand and more to make possible. We build this company to explore the frontiers of science and technology, with curiosity in our work and care for humanity and Earth.
Join us