TellWorldAI World Archive

Home›Companies

OrganizationCompany

StepFun

StepFun (阶跃星辰) is a Shanghai AI company founded in April 2023 by former Microsoft staff. It builds multimodal Step-series models and has open-sourced several of them.

Founded 2023Website stepfun.com Era Generative AI for everyone

2026 · 5 20242026
Events per year

What it is StepFun (阶跃星辰) is an AI company based in Shanghai. Its legal name is Shanghai Jieyue Xingchen Intelligent Technology Co., Ltd. It was founded on 6 April 2023 by three former Microsoft employees: Jiang Daxin, Zhu Yibo and Jiao Binxing [1]. Jiang Daxin (姜大昕) is the chief executive. Zhu Yibo leads the systems work, and Jiao Binxing, a former head of Bing's core search team, leads data [2].

The company focuses on multimodal models, which handle text, images, video and audio rather than text alone [1]. Investors have counted it among China's six so-called "AI Tiger" start-ups [1].

Origins The name comes from the step function, a mathematical function that changes value abruptly at a given point. Jiang registered the company about a month after being promoted to Microsoft vice president in March 2023 [3].

The company made its public debut on 23 March 2024 at the Global Developers Pioneer Conference in Shanghai. It announced the Step-1 language model, the Step-1V multimodal model and a preview of the trillion-parameter Step-2 model. It also launched two consumer apps, Yuewen (跃问) and Maopaoya (冒泡鸭) [1].

Main features StepFun's models are organised as a series. Step-1 and Step-1V were announced first, followed in July 2024 by the official Step-2 model, Step-1.5V and the Step-1X image generator, released at the World Artificial Intelligence Conference (WAIC, 世界人工智能大会) [4].

In 2025 the company open-sourced several models. Step-Video-T2V and Step-Audio were released in February with the carmaker Geely. Step1X-Edit followed in April, and the 321-billion-parameter Step 3 in July [1]. It also formed a model-chip alliance with Huawei (华为), Biren (壁仞科技), Moore Threads (摩尔线程) and Enflame (燧原科技) to optimise its models for domestic chips [1].

Its models reached a large installed base through phone makers. By the end of 2025 they ran on more than 42 million devices, through partnerships with about 60% of China's major phone brands. Its voice models and its AgentOS system were also used in Geely cars [5].

In 2026 the company released Step 3.5 Flash in February, Step 3.7 Flash in May and Step 5 Preview in September [1][8].

Place in the field StepFun is one of the companies that raised large sums to compete in China's model race. In January 2026 it raised a Series B+ round of more than 5 billion yuan (about US$719 million). Investors called this a record for China's AI model sector. Yin Qi, the founder of Megvii (旷视科技), became chairman, while Jiang Daxin remained chief executive [5].

On 25 February 2026 Bloomberg reported that StepFun was planning a Hong Kong initial public offering (IPO) [1][7]. In May 2026 the company completed a round of about US$2.5 billion. Investors included Huaqin Technology (华勤技术), OmniVision (豪威集团), ZTE (中兴通讯) and the Hong Kong Investment Corporation (香港投资基金公司). Earlier, in late April, StepFun had converted into a joint-stock company and dismantled its red-chip structure. At that point it had more than 500 staff [6].

20265

  1. ReleaseStepFun releases Step 5 Preview

    In September 2026 StepFun released Step 5 Preview as its new flagship foundation model. It is a sparse MoE with 600 billion total parameters, supports up to 1 million tokens of context, and is offered through the StepFun Open Platform API.

  2. ReleaseStepFun releases Step 3.7 Flash

    On 29 May 2026 StepFun released Step 3.7 Flash under the tagline "The new frontier is agent efficiency". The model adds image understanding to Step 3.5 Flash and introduces an Advisor Mode for agent tasks.

  3. FundingStepFun completes a US$2.5 billion round ahead of a Hong Kong IPO

    In May 2026 The Standard reported that StepFun had completed a funding round of about US$2.5 billion ahead of a planned Hong Kong listing, after converting into a joint-stock company in April.

  4. ReleaseStepFun releases the open Step 3.5 Flash

    In February 2026 StepFun released Step 3.5 Flash, an Apache 2.0 Mixture-of-Experts model with 196 billion total parameters, tool use and a 256K-token context. Its technical report was posted on 11 February.

  5. FundingStepFun raises a record 5 billion yuan and names Yin Qi chairman

    On 28 January 2026 StepFun announced a Series B+ funding round of more than 5 billion yuan, described as a record for China's AI model sector. Yin Qi became chairman, while Jiang Daxin remained CEO.

20255

  1. ReleaseStepFun releases the open Step 3 multimodal reasoning model

    In July 2025, during WAIC, StepFun released the open Step 3 multimodal reasoning model and formed a Model-Chip Ecosystem Innovation Alliance with Huawei, Biren, Moore Threads and Enflame.

  2. ReleaseStepFun releases Step-Audio 2

    On 23 July 2025 StepFun released the technical report for Step-Audio 2 along with two new benchmarks. Open mini versions followed on 29 August 2025.

  3. ReleaseStepFun open-sources the Step1X-Edit image editor

    On 25 April 2025 StepFun released Step1X-Edit's inference code, weights and technical report, along with GEdit-Bench, an image-editing benchmark, under Apache 2.0.

  4. ReleaseStepFun open-sources the Step-Audio speech models

    On 17 February 2025 StepFun published the Step-Audio technical report and released code and weights for its open speech models, jointly with Geely.

  5. ReleaseStepFun and Geely open-source Step-Video-T2V and Step-Audio

    On 17 February 2025 StepFun and Geely jointly open-sourced Step-Video-T2V, a 30-billion-parameter video model, and Step-Audio, a speech model family.

20242

  1. ReleaseStepFun releases Step-2, Step-1.5V and Step-1X at WAIC

    At WAIC, which opened on 5 July 2024, StepFun released the official Step-2 language model, Step-1.5V and the Step-1X image generator.

  2. ReleaseStepFun debuts with the Step-1, Step-1V and Step-2 preview models

    On 23 March 2024 StepFun made its public debut in Shanghai, announcing Step-1, Step-1V and a preview of the trillion-parameter Step-2, plus two consumer apps.

Models and products over time

12
  1. 2024
    • Mar 23, 2024 Step-1 Model

      Step-1 is StepFun's first language model, a system of roughly one hundred billion parameters unveiled in Shanghai on 23 March 2024 alongside the multimodal Step-1V and a preview of Step-2.

    • Mar 23, 2024 Step-1V Model

      Step-1V is a roughly 100-billion-parameter multimodal model from StepFun, unveiled in March 2024, that reads text, data and charts in images and supports video understanding; it ranked first on OpenCompass at launch and was succeeded by Step-1.5V.

    • Mar 23, 2024 Yuewen Product

      Yuewen is StepFun's AI assistant app, fully opened in March 2024, which reads text, data and charts in images and handles content creation, reasoning and analysis. In December 2024 its team absorbed the Maopaoya team, and the product was later renamed Step AI.

    • Jul 5, 2024 Step-2 Model

      Step-2 is a trillion-parameter language model from StepFun built on a Mixture-of-Experts architecture. A preview appeared in March 2024; the official version launched at WAIC in July 2024.

  2. 2025
    • Feb 17, 2025 Step-Video-T2V Model

      Step-Video-T2V is an open text-to-video model with 30 billion parameters. StepFun released its weights, code, a Turbo variant and a technical report under the MIT license on 17 February 2025.

    • Feb 17, 2025 Step-Audio Model

      Step-Audio is an open family of speech-interaction models released by StepFun on 17 February 2025. It includes Step-Audio-Chat, with 130 billion parameters, and Step-Audio-TTS-3B.

    • Apr 25, 2025 Step1X-Edit Model

      An open image-editing model from StepFun released in April 2025 under Apache 2.0, pairing a multimodal language model with a diffusion image decoder and shipping weights, inference code and the GEdit-Bench evaluation set.

    • Jul 23, 2025 Step-Audio 2 Model

      Step-Audio 2 is a speech model from StepFun that can understand audio and hold spoken conversations. Its mini versions were released under the Apache 2.0 license in 2025.

    • Jul 25, 2025 Step 3 Model

      Step 3 is StepFun's open multimodal reasoning model, with 321 billion total parameters and 38 billion active per token. It was released in July 2025 and uses Apache 2.0.

  3. 2026
    • Feb 2026 Step 3.5 Flash Model

      Step 3.5 Flash is an open-weights Mixture-of-Experts language model from StepFun, released in February 2026 with 196 billion total parameters, a 256K-token context and an Apache 2.0 license.

    • May 29, 2026 Step 3.7 Flash Model

      Step 3.7 Flash is a StepFun open-weights model released on 29 May 2026. It adds native image input to Step 3.5 Flash, supports a 256K context, and is licensed under Apache 2.0.

    • Sep 2026 Step 5 Preview Model

      Step 5 Preview is a preview release of StepFun's flagship foundation model: a sparse mixture-of-experts system with 600 billion total parameters, a 1-million-token context window and published weights, released in September 2026.

Notes and sources

  1. StepFunhttps://en.wikipedia.org/wiki/StepFun
  2. 阶跃星辰正式亮相,发布 Step 系列通用大模型(科创板日报)https://caifuhao.eastmoney.com/news/20240323111419028516610
  3. 阶跃星辰姜大昕与新一轮融资https://news.pedaily.cn/202406/536008.shtml
  4. WAIC 2024:阶跃星辰发布 Step-2 正式版、Step-1.5V 与 Step-1Xhttps://hub.baai.ac.cn/view/38402
  5. StepFun secures record 5 billion yuan funding, appoints new chairmanhttps://www.citynewsservice.cn/articles/shanghaidaily/news/stepfun-secures-record-5-billion-yuan-funding-appoints-new-chairman-env0pbpm
  6. Stepfun, China's AI Six Tigers, finishes new US$2.5b funding round for HK IPOhttps://www.thestandard.com.hk/finance/article/331515/Stepfun-Chinas-AI-Six-Tigers-finishes-new-US25b-funding-round-for-HK-IPO
  7. Chinese AI startup StepFun reportedly plans Hong Kong IPO, potentially raising up to $500 millionhttps://cntechpost.com/2026/02/26/stepfun-reportedly-plans-hong-kong-ipo
  8. 阶跃星辰https://zh.wikipedia.org/wiki/阶跃星辰

Cite this page: StepFun. TellWorld, https://www.tellworld.net/companies/stepfun. Updated 2026-11-08