Model Releases

StepFun Step 3.7 Flash now available on Fireworks AI

Original title:Great to see Step 3.7 Flash live on @FireworksAI_HQ. Designed for inference from day one, Step 3.7 ...

AI summary

StepFun's Step 3.7 Flash is now available on Fireworks AI. The model is a 198B sparse MoE multimodal large model (VLM) with a 196B language backbone and a 1.8B vision encoder, optimized for inference efficiency from the outset with a hardware-friendly architecture and MTP-assisted decoding, reaching 400 tokens/s. It offers native multimodal understanding and action, reliable tool use and enhanced search capabilities for real-world agent workloads, and is released under the Apache 2.0 open-source license.

Key Takeaways

  • StepFun's Step 3.
  • 7 Flash is now available on Fireworks AI.
  • The model is a 198B sparse MoE multimodal large model (VLM) with a 196B language backbone and a 1.
  • 8B vision encoder,optimized for inference efficiency from the outset with a hardware-friendly…
  • It offers native multimodal understanding and action,reliable tool use and enhanced search…

Quick Facts

Published
Updated
Source
X:阶跃星辰 StepFun (@StepFun_ai)
Category
Model Releases
AI score
77
Read the original Read the full article at the original source
Why it matters

198B稀疏MoE加MTP解码把速度推到400 tok/s,还开源Apache 2.0,这规格做agent的大脑正合适,做实时应用的可以试试手。

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: StepFun Step 3.7 Flash now available on Fireworks AI. June 4, 2026. https://new.nat6.net/en/news/cmpzitl1d03y6slkpjt6kqye1/

View all →