Today for AI
HOT RADAR
llmHEAT 10.1°

AI CURATED HOT STORY · 11/1/2026

SenseNova Open-Sources Loop-9B Model Matching 27B Performance via Looped Architecture

1 reports archived1 independent sourcesupdated 11/1/2026, 09:02:52
Synthesis & Latest Updates
1 Sources Cross-Validated

SenseNova has open-sourced the Loop-9B-Preview model, utilizing a Looped Transformer architecture to extend effective computational depth through parameter sharing. The model outperforms peers in code generation and agent tasks, validating the potential for small-parameter models to approach large-model performance via increased iterations.

LATEST/SenseNova has open-sourced the Loop-9B-Preview model, utilizing a Looped Transformer architecture to extend effective computational depth through parameter sharing. The model outperforms peers in code generation and agent tasks, validating the potential for small-parameter models to approach large-model performance via increased iterations.

Hourly heat curve

7 fully observed hours
Now
10.1
Peak
11.710:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. IT之家 · 智能时代T2·78 pts

    SenseNova Open-Sources Loop-9B Model Matching 27B Performance via Looped Architecture

    • Utilizes a Looped Transformer architecture with shared parameters in middle layers to extend computational depth while keeping original decoders at ends.
    • Introduces Manifold-Constrained Hyper-Connections (mHC) to enhance feature interaction, showing significant gains in reasoning benchmarks like GPQA Diamond.
    • Outperforms Qwen3.5-27B in specific evaluations (e.g., WorkBuddy Bench), demonstrating that a 9B model can rival larger counterparts.