Bengio’s New Parallel Multi‑Trajectory Reasoning Paradigm

The article introduces GRAM (Generative Recursive Reasoning Models), a parallel multi‑trajectory inference framework that replaces deterministic single‑track recursion with stochastic latent transitions and width scaling, achieving state‑of‑the‑art results on Sudoku‑Extreme, ARC‑AGI, N‑Queens and Graph Coloring benchmarks.

GRAMGenerative Recursive ReasoningYoshua Bengio

0 likes · 9 min read

Bengio’s New Parallel Multi‑Trajectory Reasoning Paradigm

Machine Learning Algorithms & Natural Language Processing

May 23, 2026 · Artificial Intelligence

10M‑Parameter Model Solves ARC and Sudoku – Bengio Team Bets on Multi‑Trajectory Reasoning

A 10‑million‑parameter GRAM model from Bengio, KAIST, Mila and NYU achieves 97% accuracy on Sudoku‑Extreme and competitive scores on ARC‑AGI tasks by replacing deterministic recursive updates with a probabilistic multi‑trajectory process, and extensive ablations show that both random guidance and depth‑supervised training are essential for its performance.

ARC‑AGIGRAMGenerative Recursive Reasoning

0 likes · 9 min read

10M‑Parameter Model Solves ARC and Sudoku – Bengio Team Bets on Multi‑Trajectory Reasoning

Machine Heart

May 23, 2026 · Artificial Intelligence

Bengio’s New Paper Pushes Recursive Reasoning Limits with Parallel Trajectories

The paper introduces GRAM (Generative Recursive Reasoning Models), a probabilistic multi‑trajectory recursive reasoning framework that injects learnable randomness into each recursion step, enabling parallel sampling and achieving higher accuracy than deterministic baselines across tasks such as Sudoku‑Extreme, N‑Queens, ARC‑AGI and unconditional generation.

AI modelsGRAMparallel sampling

0 likes · 12 min read

Bengio’s New Paper Pushes Recursive Reasoning Limits with Parallel Trajectories

Machine Learning Algorithms & Natural Language Processing

May 22, 2026 · Artificial Intelligence

How a 10M‑Parameter Model Beats Large Models on Sudoku and ARC with Multi‑Trajectory Reasoning

The GRAM model introduced by Yoshua Bengio’s team replaces deterministic recursive updates with probabilistic multi‑trajectory sampling, enabling a 10 M‑parameter network to achieve 97 % accuracy on Sudoku‑Extreme, 52 %/11 % on ARC‑AGI, and near‑perfect results on N‑Queens and graph‑coloring, while also supporting unconditional generation tasks.

ARC‑AGIGRAMSudoku

0 likes · 9 min read

How a 10M‑Parameter Model Beats Large Models on Sudoku and ARC with Multi‑Trajectory Reasoning