WhoBuilt

Hierarchical Reasoning Model

Also known as HRM, HRMs, hierarchical reasoning models

1 Talk3 Facts3 Sources

Key facts

Capabilities
Uses fixed-point iteration with stop-gradient instead of backpropagating through all recursion steps
Uses low-level, high-level, and outer refinement recursion

From talks

Results and examples shared in single talks

Timeline 2

Dated events, newest first. Open the sources under any item.

How we verify
20262 events
  1. as of May 2026Capability
    Uses fixed-point iteration with stop-gradient instead of backpropagating through all recursion steps
    “What they do is use this DEQ method of fixed-point iteration.” 11:22
    Unverified1 source· Talk
  2. as of May 2026Capability
    Uses low-level, high-level, and outer refinement recursion
    “There are exactly three levels of recursion: the low level, the high level, and the outer refinement steps.” 8:22
    Unverified1 source· Talk

Talks 1

37:53
Recursion Is The Next Scaling Law In AI
François Chollet, Wa C

The talk explains how recursion can give AI models more reasoning depth at inference time without requiring more parameters. It compares hierarchical reasoning models (HRM) with tiny recursive models (TRM), highlighting repeated refinement of latent states and truncated backpropagation as key ideas. The speakers argue that these methods can solve tasks such as Sudoku and ARC with small, task-specific models, without step-by-step reasoning traces. They close by suggesting that recursive reasoning could complement the broad representations learned by large language models.

May 1, 2026· 38 minDiscussed 0:00