{
  "id": "stepfun",
  "title": "StepFun",
  "lang": "en",
  "source_lang": "en",
  "summary": "Chinese artificial intelligence company based in Shanghai, known for the multimodal Step series of foundation models.",
  "category": "computer-science",
  "subcategory": "artificial-intelligence",
  "type": "organization",
  "aliases": [
    "阶跃星辰",
    "Shanghai Jieyue Xingchen Intelligent Technology Co., Ltd"
  ],
  "generated_at": "2026-10-16",
  "model": "pro/claude-haiku-5-5",
  "url": "https://aiwiki.page/en/stepfun/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/stepfun/",
      "json": "https://aiwiki.page/zh/stepfun.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "StepFun (Chinese: 阶跃星辰) is a artificial intelligence company headquartered in Shanghai, China. Its legal name is Shanghai Jieyue Xingchen Intelligent Technology Co., Ltd. It was founded in April 2023 by former Microsoft employees, with Jiang Daxin, Zhu Yibo and Jiao Binxing listed among its founders. The company develops large language models and other foundation models under the Step brand, and it is commonly grouped with other well-funded Chinese AI start-ups that investors have nicknamed the \"AI Tiger\" companies. The firm's stated focus is general-purpose AI, and from the outset it has emphasised multimodal capability rather than text alone."
    },
    {
      "heading": "History",
      "text": "The company was registered in April 2023. Within two months, it trained its first 100-billion-parameter model, Step 1, on its first attempt. In July 2024, at the World Artificial Intelligence Conference, StepFun officially launched Step-2, a trillion-parameter LLM, along with the Step-1.5V multimodal model and the Step-1X image generation model. Investors cited the July 2024 releases as a reason for confidence in the company.\n\nIn 2025 the company continued to release models and open weights. In February 2025, StepFun and Geely jointly announced the open-sourcing of two multimodal large models to global developers, Step-Video-T2V and Step-Audio. In July 2025, StepFun released Step 3. The Model-Chip Ecosystem Innovation Alliance aimed to optimize Step 3 for domestic chips. Step-R1-V-Mini, released in April 2025, is a multimodal reasoning model designed for visual interpretation and image understanding.\n\nIn early 2026 the company released its most prominent open model. In February 2026, Step 3.5 Flash, a mixture-of-experts model with 196 billion parameters and 11 billion active parameters, was released under the Apache 2.0 license. It supports tool use and a 256k token context window. Another source gives the total parameter count as 198B total parameters with 11B active parameters, so the exact figure should be treated with caution. Later releases listed in one reference table include Step 3.7 Flash in May 2026 under Apache 2.0, and Step 5 Preview in September 2026 with closed weights."
    },
    {
      "heading": "Products and model lineup",
      "text": "The company's model family is referred to collectively as the Step series. According to a Chinese-language encyclopedia entry, the series spans text, multimodal, image generation, video generation and voice interaction models. The company also operates a consumer chat application, which has been renamed over time, and it has pursued agent-oriented products. Step 5 Preview is described as a larger general-purpose model designed to power agents in the cloud, and a press report on its launch emphasised low per-token pricing."
    },
    {
      "heading": "Funding and ownership",
      "text": "Reported funding figures differ considerably between sources, largely because of differences in how rounds and currencies are counted. A January 2026 round was reported as more than $718M in a Series B+ round announced on January 26th, according to TechNode. New participants included Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital and others, while Tencent, Qiming Venture Partners and 5Y Capital invested again. Fortera Capital, the private-equity arm of Shanghai State-owned Capital Investment, led the earlier Series B round. A PitchBook profile gives a much higher cumulative figure of $3.07B, while a general-purpose tool directory estimates ~$1B+ raised. The January 2026 round was earmarked for foundation-model development and StepFun's strategy of putting AI into phones, cars and other devices.\n\nCommentators have described the company's market position in terms of consolidation. Z.ai and MiniMax listed in Hong Kong, while StepFun and Moonshot secured large financing rounds."
    },
    {
      "heading": "Strategy and positioning",
      "text": "StepFun has adopted a mixed release strategy. One profile states that the company releases some models as open weights and by 2026 was shipping trillion-parameter mixture-of-experts systems, though it remains smaller than China's largest AI developers. Its commercial plans combine model APIs with device-level integration, particularly with smartphone makers and automotive partners. An internal self-description reported by a business publication compared the company to a Chinese counterpart of xAI + Tesla, although this is a secondhand characterisation."
    },
    {
      "heading": "Controversies and open questions",
      "text": "In September 2026, a timeline aggregator reported that a joint NSA/CISA/FBI advisory named DeepSeek, Moonshot, Alibaba, MiniMax, StepFun and Z.AI as running systematic distillation campaigns against Claude, GPT, Gemini and Grok since late 2024; China's Commerce Ministry rejected it and threatened countermeasures. The report is not corroborated by the primary advisory text in the sources consulted, and the claim should be regarded as unverified.\n\nIndependent verification of company performance is limited. Benchmark claims published by the company and by third parties vary in methodology, and the company's own reported figures for user numbers and device installations could not be confirmed against independent sources."
    }
  ],
  "links": [
    "artificial-intelligence",
    "shanghai",
    "china",
    "large-language-model",
    "multimodal-learning",
    "ai-agent",
    "smartphone",
    "electric-vehicle"
  ],
  "sources": [
    {
      "title": "StepFun",
      "url": "https://en.wikipedia.org/wiki/StepFun"
    },
    {
      "title": "As AI consolidates, what makes StepFun worth a RMB 5 billion raise?",
      "url": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise"
    },
    {
      "title": "阶跃星辰（StepFun）- 面向智能体时代的AI大模型",
      "url": "https://www.navcn.cn/tool/2401.html"
    },
    {
      "title": "阶跃星辰 - 维基百科，自由的百科全书",
      "url": "https://zh.wikipedia.org/wiki/%E9%98%B6%E8%B7%83%E6%98%9F%E8%BE%B0"
    },
    {
      "title": "StepFun launches a 600B agent model at $1 per million input tokens",
      "url": "https://runtimewire.com/article/stepfun-step-5-preview-600b-agent-model-pricing"
    },
    {
      "title": "Stepfun 2026 Company Profile: Valuation, Funding & Investors",
      "url": "https://pitchbook.com/profiles/company/721081-18"
    },
    {
      "title": "Stepfun Review (2026): Chinese AI Frontier Lab",
      "url": "https://tooldirectory.ai/tools/stepfun"
    },
    {
      "title": "StepFun: AI timeline & profile • It Does What Now?",
      "url": "https://itdoeswhatnow.com/tag/stepfun/"
    },
    {
      "title": "AI \"Step\" created by StepFun — A former Microsoft Chief Scientist aims for the \"pinnacle of scaling\"｜香川友志",
      "url": "https://note.com/kagawatomo/n/n84e631eb3252?hl=en"
    }
  ],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}