Skip to main content

HiDream.ai

Category: Foundation Models / LLMs

HiDream.ai (智象未来) is a Chinese AI startup building visual multimodal foundation models for image, video, text, and 3D generation, with proprietary native full-modal architecture (UiT). HiDream.ai was founded in 2023. The company is led by 梅涛 (Tao Mei). Based in Beijing, China. Team size: 51-200. Total funding raised: $83.0M. Latest round: Series B. Key investors include 深创投 (Shenzhen Capital); 东方富海 (Oriental Fortune Capital); 金浦投资 (Jinpu Capital); 京东集团 (JD.com); 安徽省投资集团/安徽省产投; 合肥产投; 科大讯飞 (iFlytek); 峰华资本 (Fenghua Capital); 敦鸿资本 (Dunhong Capital); 阿尔法公社 (Alpha Startup Fund); 将门创投 (Jiangmen VC).

Founded
2023
Headquarters
Beijing, China
Team size
51-200
Total funding
$83.0M

Value proposition

Native full-modal (原生全模态) visual generation foundation model covering text, image, video, and 3D simultaneously under one unified architecture (Unified Transformer / UiT), achieving SOTA on multiple benchmarks while being the first Chinese generative AI model to top the global Artificial Analysis leaderboard.

Products and solutions

HiDream series foundation models (HiDream-O1-Image-Pro 200B+ closed-source, HiDream-O1-Image 8B open-source, HiDream-I1 17B open-source MIT); Pixeling (千象) design tool; PixMaker marketing content platform; vivago.ai social media agent app (40M+ users globally); HiBurst e-commerce marketing agent (TikTok top 5 service provider); 帧赞 (ZhenZan) professional AI film production agent; HiHarness enterprise capability platform

Unique value

Only global base model simultaneously supporting four modalities (text, image, video, 3D) with a native full-modal architecture (UiT) — not a patchwork of single-modal models. Released 200B+ parameter HiDream-O1-Image-Pro, the world's largest native multimodal image model.

Target customer

E-commerce marketers (cross-border TikTok/Douyin), professional film/video production studios, social media content creators, designers, brand marketing teams

Industries served

E-commerce & cross-border marketing; Film & TV production; Social media content creation; Advertising & brand marketing; Visual design

Technology advantage

Proprietary Unified Transformer (UiT) native full-modal architecture — unified encoding of pixels, text tokens, and task conditions in shared continuous latent space. HiDream-O1-Image-Pro at 200B+ parameters is the world's largest native multimodal image model. First Chinese generative AI model to top Artificial Analysis global leaderboard (HiDream-O1-Image 8B open-source model ranked #1 globally). Models iterate every 2 weeks.

How they differentiate

Native full-modal (UiT) architecture vs. traditional fragmented multi-module approaches (U-Net + separate text encoders). All-in-one text-image-video-3D under one architecture. First to ship text-to-video globally. Open-source strategy (HiDream-I1 under MIT license, HiDream-O1-Image open-source) while offering massive closed-source enterprise models. TikTok top 5 service provider for e-commerce video generation.

Main competitors

Midjourney (global); OpenAI DALL-E 3 / Sora (global); ByteDance Jimeng (China); Kuaishou Kling (China)

Key partnerships

寒武纪 (Cambricon) — strategic partnership & model adaptation; 华为云 (Huawei Cloud); 腾讯云 (Tencent Cloud); 上海电影集团上影新视野基金 (Shanghai Film Group); 蓝色光标 (BlueFocus); 北京捷成世纪; 倍尔健康; 诺亦腾机器人 (Noitom Robotics) — embodied AI data; 联想 (Lenovo); 中国移动 (China Mobile); 微软中国 (Microsoft China); TikTok — top 5 official service provider

Notable customers

TikTok (top 5 official service provider, 1M+ e-commerce marketing videos produced); 4000万+ users across 100+ countries on vivago.ai; 1000+ professional filmmaking teams on 帧赞 platform; 50+ enterprise partners including Lenovo, China Mobile, Microsoft China

Major milestones

2023-03: Company founded; 2023-04: Seed funding from Alpha Startup Fund; 2023-09: Launched 智象视觉多模态生成式大模型 (first multimodal model); 2023-12: Passed model & algorithm dual registration; Angel round led by iFlytek; 2024-12: Pre-A and A rounds totaling several hundred million RMB; 2025-07: Open-sourced HiDream-I1 (17B, MIT), first Chinese model to top Artificial Analysis global leaderboard; 2025-12: A+ round from JD.com; 2026-04: B round >500M RMB; 2026-05: Released HiDream-O1-Image-Pro (200B+ params); B+ round from Shenzhen Capital etc; CEO Mei Tao elected ACM Fellow (2025)

Growth metrics

As of Q1 2026: 30M+ professional users globally across vivago.ai, 40K+ enterprise customers, Q1 2026 revenue exceeded full-year 2025 revenue, HiDream-I1 open-source model downloads exceeded 2M. Products cover 100+ countries.

Market positioning

Leading Chinese generative AI company in visual multimodal foundation models. Competes globally with Midjourney and DALL-E 3 on image quality, while targeting the Chinese and Asian e-commerce, marketing, and film production markets with localized products.

Geographic focus

China (primary), global markets (vivago.ai reaches 100+ countries)

Patents and IP

Unified Transformer (UiT) native full-modal architecture (proprietary); Multiple papers at CVPR 2024, ICML 2025, ACM MM 2025

About 梅涛 (Tao Mei)

Ex-JD.com Vice President; Ex-Microsoft Research Senior Researcher; Ph.D. from University of Science and Technology of China. IEEE Fellow, IAPR Fellow, CAAI Fellow, Canadian Academy of Engineering International Fellow, ACM Fellow (2025). 50,000+ academic citations.

Official website: