🛰️ Daily AI Frontier
‹ back to 2026-09-20

阶跃新旗舰 Step 5 Preview 发布,跻身 AA 榜单全球开源前三

Industry & News LLMs & Foundation Models

Ranking

Overall 75
Content 85
Popularity N/A

No observed public metrics; popularity remains neutral/archived.

Representative image for 阶跃新旗舰 Step 5 Preview 发布,跻身 AA 榜单全球开源前三

Merged summary

TL;DR - StepFun announced Step 5 Preview, a multimodal sparse-MoE foundation model targeting agentic software engineering, professional, and financial workflows. It ranks among the top three open models on Artificial Analysis’s Intelligence Index and is scheduled for open-source release on October 15.

  • Uses 600B total parameters with 27B active parameters per inference and supports a 1-million-token context window.
  • Natively accepts text and visual inputs, with an emphasis on complex agent tasks and long-horizon planning.
  • StepFun reports a per-task cost one-eighth that of Claude Opus 5, positioning efficiency as a key advantage.
  • Public and internal evaluations reportedly show strong performance in software engineering, professional knowledge work, and finance.

Sources (1)

阶跃新旗舰 Step 5 Preview 发布,跻身 AA 榜单全球开源前三

雷峰网 (AI科技评论) 2026-09-20
Public signals N/A
Providers: Hugging Face · N/A OpenAlex · N/A Publisher · N/A Semantic Scholar · N/A X · N/A Fetched 2026-09-26 14:15:03.571564 UTC

TL;DR - StepFun announced Step 5 Preview, a multimodal sparse-MoE foundation model targeting agentic software engineering, professional, and financial workflows. It ranks among the top three open models on Artificial Analysis’s Intelligence Index and is scheduled for open-source release on October 15.

  • Uses 600B total parameters with 27B active parameters per inference and supports a 1-million-token context window.
  • Natively accepts text and visual inputs, with an emphasis on complex agent tasks and long-horizon planning.
  • StepFun reports a per-task cost one-eighth that of Claude Opus 5, positioning efficiency as a key advantage.
  • Public and internal evaluations reportedly show strong performance in software engineering, professional knowledge work, and finance.
item →