←Home

    How Models Learn to Reason

    A guided tour of current reasoning research: how reinforcement learning, distillation, verification, test-time compute and latent computation produce machine reasoning, and what still blocks it.

    In order

    1 stories
    surreal render: a person sits on a ledge in a void of hanging screens and cables
    surreal render: a person sits on a ledge in a void of hanging screens and cables

    Surreal render: a person sits on a ledge in a void of hanging screens and cables · Photo: Shubham Dhage

    Agency · Olympiad reasoning

    A compact recipe and heavy self-checking take a mid-sized model to olympiad gold level

    Given a compact training recipe, an existing reasoning model scored 35 of 42 on IMO 2025 and USAMO 2026 problems, but only with a long, costly draft-and-check loop.

    arXiv preprint · 13 May 2026