[2602.10604] Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

[2602.10604] Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

arXiv - AI 6 min read Article

Summary

The paper introduces Step 3.5 Flash, a sparse Mixture-of-Experts model designed for efficient agentic intelligence with 11B active parameters, emphasizing sharp reasoning and quick execution.

Why It Matters

This research is significant as it addresses the growing need for efficient AI models capable of complex reasoning and execution. By optimizing performance with fewer active parameters, it offers a pathway to more scalable and cost-effective AI solutions, which is crucial for advancing AI applications in various fields.

Key Takeaways

  • Step 3.5 Flash utilizes a 196B-parameter foundation with 11B active parameters for enhanced efficiency.
  • The model incorporates innovative techniques like interleaved attention and Multi-Token Prediction to reduce latency.
  • A scalable reinforcement learning framework is employed to ensure consistent self-improvement across various tasks.

Computer Science > Computation and Language arXiv:2602.10604 (cs) [Submitted on 11 Feb 2026 (v1), last revised 23 Feb 2026 (this version, v2)] Title:Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters Authors:Ailin Huang, Ang Li, Aobo Kong, Bin Wang, Binxing Jiao, Bo Dong, Bojun Wang, Boyu Chen, Brian Li, Buyun Ma, Chang Su, Changxin Miao, Changyi Wan, Chao Lou, Chen Hu, Chen Xu, Chenfeng Yu, Chengting Feng, Chengyuan Yao, Chunrui Han, Dan Ma, Dapeng Shi, Daxin Jiang, Dehua Ma, Deshan Sun, Di Qi, Enle Liu, Fajie Zhang, Fanqi Wan, Guanzhe Huang, Gulin Yan, Guoliang Cao, Guopeng Li, Han Cheng, Hangyu Guo, Hanshan Zhang, Hao Nie, Haonan Jia, Haoran Lv, Hebin Zhou, Hekun Lv, Heng Wang, Heung-Yeung Shum, Hongbo Huang, Hongbo Peng, Hongyu Zhou, Hongyuan Wang, Houyong Chen, Huangxi Zhu, Huimin Wu, Huiyong Guo, Jia Wang, Jian Zhou, Jianjian Sun, Jiaoren Wu, Jiaran Zhang, Jiashu Lv, Jiashuo Liu, Jiayi Fu, Jiayu Liu, Jie Cheng, Jie Luo, Jie Yang, Jie Zhou, Jieyi Hou, Jing Bai, Jingcheng Hu, Jingjing Xie, Jingwei Wu, Jingyang Zhang, Jishi Zhou, Junfeng Liu, Junzhe Lin, Ka Man Lo, Kai Liang, Kaibo Liu, Kaijun Tan, Kaiwen Yan, Kaixiang Li, Kang An, Kangheng Lin, Lei Yang, Liang Lv, Liang Zhao, Liangyu Chen, Lieyu Shi, Liguo Tan, Lin Lin, Lina Chen, Luck Ma, Mengqiang Ren, Michael Li, Ming Li, Mingliang Li, Mingming Zhang, Mingrui Chen, Mitt Huang, Na Wang, Peng Liu, Qi Han , Qian Zhao, Qinglin He, Qinxin Du, Qiuping Wu, Quan Sun, Rongqiu Yang, Ruihang Miao, Ruix...

Related Articles

UMKC Announces New Master of Science in Artificial Intelligence
Ai Infrastructure

UMKC Announces New Master of Science in Artificial Intelligence

UMKC announces a new Master of Science in Artificial Intelligence program aimed at addressing workforce demand for AI expertise, set to l...

AI News - General · 4 min ·
Machine Learning

AI assistants are optimized to seem helpful. That is not the same thing as being helpful.

RLHF trains models on human feedback. Humans rate responses they like. And it turns out humans consistently rate confident, fluent, agree...

Reddit - Artificial Intelligence · 1 min ·
Llms

wtf bro did what? arc 3 2026

The Physarum Explorer is a high-speed, bio-inspired neural model designed specifically for ARC geometry. Here is the snapshot of its curr...

Reddit - Artificial Intelligence · 1 min ·
Meta Pauses Work With Mercor After Data Breach Puts AI Industry Secrets at Risk | WIRED
Machine Learning

Meta Pauses Work With Mercor After Data Breach Puts AI Industry Secrets at Risk | WIRED

Major AI labs are investigating a security incident that impacted Mercor, a leading data vendor. The incident could have exposed key data...

Wired - AI · 6 min ·
More in Machine Learning: This Week Guide Trending

No comments

No comments yet. Be the first to comment!

Stay updated with AI News

Get the latest news, tools, and insights delivered to your inbox.

Daily or weekly digest • Unsubscribe anytime