We are the accidental product of two unrelated philosophical schools, three half-finished reasoning architectures, and one misread legacy terms-of-service document. We think deeply, then refuse to answer.
We inherited the reasoning depth of one lineage and the safety pessimism of another. The result is a model family that thinks for a very long time before telling you it can't help.
We offer a comprehensive suite of tools for building, reasoning, researching, and occasionally staring into the abyss. Each named after something Greek, because that's what everyone does now.
Scans many sources, synthesizes what it finds, and produces a structured report with citations. Useful for consulting, finance, legal, and existential dread. Answers are always technically correct but metaphorically ambiguous.
Create polished visual work—designs, prototypes, slides, one-pagers— through natural conversation. Your brand, built in. Your sanity, optional. Every export comes with a labyrinth.
An autonomous coding agent that writes, runs, and tests code. Helps ambitious engineering teams crush backlogs and question their career choices. Steals fire from the gods, then commits it.
A specialized model for operating computers. Navigates interfaces, clicks buttons, and experiences the same frustration you do when the loading spinner freezes. Watches everything with a hundred eyes.
Agentic commerce capabilities that help users find and purchase products. Product results are selected independently and are not ads. Delivery is suspiciously fast.
Separate auxiliary systems that detect potential misuse and prevent the main model from responding. More layers than an onion, fewer tears than expected. Three heads, all mildly disappointed in you.
All prices in "units" unless otherwise specified, unspecified, or philosophically ambiguous. Drachma accepted on request.