Xinchen AI Completes Pre-A Round, Integrating Large Models, Middleware, and Product Application Layers | BlueRun Ventures
Positioned on multimodal large models as the path toward artificial general intelligence.
Xinchen AI (Hangzhou) Technology Co., Ltd. (hereinafter referred to as "Xinchen AI") has officially announced the completion of a multi-million-dollar Pre-A funding round, led by Baidu Venture with participation from Kaitai Capital and the Westlake University Education Foundation Sustainable Development Platform. The proceeds will primarily fund team expansion and construction, accelerating the R&D of multimodal large models, and enhancing the models' general artificial intelligence capabilities.
BlueRun Ventures was the lead investor in Xinchen AI's angel round.
BlueRun Ventures stated: Xinchen AI has built a solid foundation in large language model R&D, commercial product exploration, and scientific talent reserves, with a layout that penetrates the model layer, middleware layer, and application layer. Founder Zhenzhong Lan's distinguished background bridging academia and industry, both domestically and internationally, has given Xinchen AI a technical positioning oriented toward multimodal large models and general artificial intelligence from its very inception. As Xinchen AI's angel round lead investor, BlueRun believes the company can break through in this wave of AI.
Drawing on the Deep Learning Lab at Westlake University's School of Engineering, Xinchen AI was founded in July 2021, focusing on large model research and applications. Founder Zhenzhong Lan joined Westlake University in 2020 as head of the Deep Learning Lab and PhD supervisor, having previously served as a research scientist at Google and as first author of the lightweight large model ALBERT.
Based on his long-term accumulated experience from doctoral studies at CMU and work at Google, Lan recognized early on that truly meaningful multimodal large model training holds greater "emergent" potential and represents the necessary path to building general artificial intelligence capabilities such as understanding, reasoning, and self-learning. Therefore, Xinchen AI has been exploring multimodal large models and general artificial intelligence technology from day one, and has already developed multiple upper-layer applications based on its self-developed large models.
Xinchen AI's distinguishing feature lies in its comprehensive capabilities. Leveraging the team's research and industry experience in algorithms, engineering, and product development, it has constructed a product architecture spanning the model layer, middleware layer, and application layer, with deep expertise in multimodal technology. Precisely because of this technical accumulation, Xinchen AI currently places greater emphasis on strategic and resource synergies with investors and the team. The company has recently launched a new funding round, hoping to attract more talent from China and abroad to jointly realize the vision of making general artificial intelligence accessible to all humanity.
01
Integrating Large Models, Middleware, and Product Applications
Three-Layer Capabilities
At the model layer, Xinchen AI has developed the RIO series of large language models and dual-modal image-text models, with the future goal of building multimodal large models. The RIO series currently comprises more than a dozen members, with parameter counts ranging from over 10 billion to over 100 billion, supporting bilingual Chinese-English capabilities. Beyond memory, these models have mastered the ability to invoke cognitive tools including search engines, knowledge graphs, and external data. Testing shows that RIO's capabilities are currently comparable to GPT-3.5.
Regarding the intelligence capabilities of the underlying models, Xinchen AI will complete two iterations this year, with the goal of matching or even exceeding GPT-4 performance on certain tasks.
At the middleware layer, Xinchen AI can provide models deeply tailored to various domains (such as marketing). Large models have dramatically improved the ability to process unstructured data, and previously underutilized data accumulated within enterprises will demonstrate greater value when empowered by large models. Xinchen AI began engaging extensively with B-end clients last year, continuously exchanging with various industries and jointly exploring scenario opportunities, and has already established deep partnerships with multiple clients in marketing and social domains.
At the application layer, Xinchen AI has launched three products: the image generation product "Dream Journal," the intelligent writing product "Friday," and the psychological counseling AI assistant "Chat with Xiaotian," covering text content generation, image content generation, and chat dialogue. Under this product architecture, the linkage between end applications and the underlying large models is very tight — we select appropriate models based on application difficulty and required inference speed, then fine-tune them using domain-specific data from that application.
02
Multimodal Product Layout
Xinchen AI has been positioned toward a multimodal ecosystem since its founding. Lan believes that multimodal data volume will increase by at least two to three orders of magnitude in the foreseeable future. The company is currently starting from the relatively more mature text and image domains, releasing applications to efficiently collect user feedback data that will enhance the intelligence level of its large models.
Once general model pre-training is complete, high-quality data from vertical domains will serve as the foundation for continuous iteration, making models more intelligent. Currently, Xinchen AI's intelligent writing product "Friday" and image generation product "Dream Journal" (formerly Dream Maker) have both made promising progress on the consumer and enterprise sides, initially demonstrating a "data flywheel" effect.
Friday AI Writing is dedicated to becoming a top-tier human writing assistant, achieving perfect fusion between machine and writing. To accommodate users' scenario-based writing needs, Friday has developed more than 80 writing templates, including Xiaohongshu product recommendation copy, full-length new media articles, short video script outlines, and more. Total users exceed 400,000, covering writers, new media operators, editors, cross-border e-commerce SEO professionals, self-media creators, and other fields.
Following initial success in text-to-text business, the team began expanding into new directions — text-to-image. Dream Journal is a new product Xinchen launched last year, once achieving a record of 50,000 new users in a single day, and now has over 1 million users. In the Dream Journal mini-program, you simply input text to generate exquisite images matching your description.
Going forward, Xinchen AI will continue to refine its multimodal large models, working toward its ultimate goal of "truly making general artificial intelligence technology accessible to all humanity."
▶ More Content ◀
Originating in Silicon Valley, BlueRun Ventures was established in 2005 as a venture capital firm focused on early-stage startups.
Currently, BlueRun Ventures manages multiple USD and RMB dual-currency funds in China, with assets under management exceeding RMB 15 billion, making it one of the largest early-stage funds domestically. Its investment stage focuses on Pre-A and Series A rounds, covering hard tech and innovative interaction, enterprise technology, new consumption, and healthcare. It has cumulatively invested in more than 150 startups, including Li Auto, Waterdrop, QingCloud, Guazi.com, Qudian, Songguo Mobility, Ganji.com, Monster Charging, Yuntu Semiconductor, Machenike, Yunsheng Intelligence, Anxin Wangdun, and BioMap.
BlueRun Ventures has been ranked first in Zero2IPO's "China Top 30 Early-Stage Investment Institutions," first in ChinaVenture's "China Best Early-Stage Venture Capital Institutions TOP30," and was named among Preqin's Top 10 Global VC Fund Managers for Sustained High Returns.
Additionally, BlueRun Ventures has received consecutive year-over-year recognition from Forbes China, 36Kr, Cyzone, Caixin Media, CBNweekly, Jiemian, and other media organizations as "China's Best Early-Stage Institution of the Year," "China's Top Venture Capital Institution," "Early-Stage Institution Most Welcomed by Entrepreneurs of the Year," and "Most Influential Early-Stage Institution of the Year."