<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>StepFun (Step-Video &amp; Step-Audio) — FoundationIndex</title>
    <link>https://foundationindex.ustunozgur.com/s/stepfun</link>
    <description>The Shanghai AI lab founded by former Microsoft Research Asia Chief Scientist Jiang Daxin—known for open-sourcing the 30B Step-Video-T2V generator, the 130B Step-Audio voice model, Step1X-Edit image editor, and the hardware-co-designed Step-3 multimodal MoE.</description>
    <language>en</language>
    <atom:link href="https://foundationindex.ustunozgur.com/s/stepfun/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>StepFun (Step-Video &amp; Step-Audio) · Paper · 2025-07-25</title>
      <link>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step3-mfa-afd-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step3-mfa-afd-2025</guid>
      <pubDate>Fri, 25 Jul 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;StepFun published Step-3 (321B total / 38B active MoE VLM), introducing Multi-Matrix Factorization Attention (MFA) to cut KV cache traffic and Attention-FFN Disaggregation (AFD) to pipeline attention and expert layers across separate GPU pools.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2507.19427&quot;&gt;Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decoding&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>StepFun (Step-Video &amp; Step-Audio) · Release · 2025-04-24</title>
      <link>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step1x-edit-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step1x-edit-2025</guid>
      <pubDate>Thu, 24 Apr 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;StepFun open-sourced Step1X-Edit, coupling a Qwen-VL multimodal encoder with a DiT image denoiser to perform natural-language reference image editing locally.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2504.17761&quot;&gt;Step1X-Edit: A Practical Framework for General Image Editing&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>StepFun (Step-Video &amp; Step-Audio) · Paper · 2025-03-18</title>
      <link>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step-video-ti2v-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step-video-ti2v-2025</guid>
      <pubDate>Tue, 18 Mar 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;StepFun released Step-Video-TI2V (30B), extending its open-weight video foundation backbone to text-conditioned image-to-video generation with explicit motion-magnitude scoring and camera-stability control.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2503.11251&quot;&gt;Step-Video-TI2V Technical Report: A State-of-the-Art Text-Driven Image-to-Video Generation Model&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>StepFun (Step-Video &amp; Step-Audio) · Release · 2025-02-14</title>
      <link>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step-video-audio-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/stepfun#stepfun-step-video-audio-2025</guid>
      <pubDate>Fri, 14 Feb 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;StepFun open-sourced Step-Video-T2V—a 30B-parameter DiT trained with a 16x16x8 Video-VAE and Video-DPO to generate 204-frame clips—alongside Step-Audio, a 130B open-weight speech and voice interaction framework.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2502.10248&quot;&gt;Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Model&lt;/a&gt;, &lt;a href=&quot;https://arxiv.org/abs/2502.11946&quot;&gt;Step-Audio: Unified Understanding and Generation in Intelligent Speech Interaction&lt;/a&gt;&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
