HRM-Text - sapient.inc

Skip to contentTry HRM<br>HRM-Text<br>Open-sourced in May 2026, HRM-Text is a 1B text generation model based on the HRM architecture, strengthened by task completion and latent space reasoning.<br>Download HRM-Text

Key Traits<br>Data-Efficient Training<br>Trained on ~40B tokens, using up to 1000× less data than the 4–36T tokens used by the models we benchmark against.

Compact Yet Powerful<br>Built with 1.15B parameters while remaining competitive with models several times its size on reasoning-heavy benchmarks.

Native Edge Reasoning<br>Runs locally with a 0.6 GiB footprint at int4 quantization, enabling advanced reasoning without cloud dependency.

Application Domains<br>Our architecture powers advanced reasoning across complex, high-impact real-world domains.

Embodied AI

Healthcare

Quantitative Finance

Climate

AI4S

Benchmarks<br>HRM-Text is a proof-of-concept model with no post-training. The numbers below reflect architecture performance alone.

MATH

DROP

ARC-C

MMLU

Despite its compact size, HRM-Text delivers competitive results across reasoning benchmarks, including 56.2% on MATH, 81.9% on ARC-Challenge, 82.2% on DROP, and 60.7% on MMLU.

Benchmark Explanations<br>MATH:<br>A benchmark that tests mathematical reasoning and problem solving, often requiring multi-step logic rather than simple recall.

ARC-C:<br>The AI2 Reasoning Challenge- Challenge Set, designed to test science reasoning through difficult grade-school science questions that require inference and commonsense understanding.

DROP:<br>A reading comprehension benchmark that tests a model’s ability to reason over passages, especially with numbers, counting, comparison, and discrete operations.

MMLU<br>Massive Multitask Language Understanding, a broad benchmark covering many subjects, used to evaluate general knowledge and multi-domain reasoning.

Explore HRM-Text on GitHub

View Repository

Copyright © 2026. All rights reserved.<br>Privacy PolicyTerms & ConditionsCookie Notice