Hongxiong AI Secures Hundreds of Millions in Series A+ Funding for "Memory Science"
Hongxiong AI has announced the completion of a Series A+ funding round, raising hundreds of millions of RMB and achieving a post-investment valuation of nearly 3 billion RMB. This round was led by Zhejiang Jiuwei Private Equity Fund, Jiaxing Zhangyuan Venture Capital, and existing shareholder Greyfeng. This marks the sixth funding round for Hongxiong AI within a span of just 15 months. The raised capital will primarily fuel advancements in foundational research for AI "memory science" and its human-like brain principles, accelerate the deep integration of its OpenBear general large model with the MemoryBear memory science system, and expand market reach across four core application areas: intelligent customer service, intelligent marketing, ChatBI, and intelligent education.
The company's strategy centers on "memory-driven AGI" and addresses the limitations of traditional large language models, such as token limits causing context loss, high knowledge forgetting rates, and increased operational costs. Hongxiong AI's MemoryBear system, inspired by human cognitive mechanisms, incorporates layered dynamic memory (working, short-term, and long-term), time memory technology, dynamic semantic networks, intelligent semantic pruning, a self-reflection engine, and minimized memory sharing. This system reportedly reduces knowledge forgetting rate to below 1%, decreases token consumption by 25 times, and lowers the hallucination rate to approximately 0.2%.
Hongxiong AI's core technological ecosystem comprises the OpenBear large model, the MemoryBear memory science system, and a native Agent architecture. OpenBear is designed with memory interfaces for seamless interaction with MemoryBear. The MemoryBear system employs theories like ACT-R, the Ebbinghaus forgetting curve, and a 3D self-reflection engine to manage information dynamically. The native Agent architecture is built with inherent memory capabilities, enabling personalized interactions and autonomous task execution. The company is also extending its capabilities to the consumer market with the upcoming launch of the OpenBear general large model for consumers and the CodeBear memory-assisted programming tool. Commercially, Hongxiong AI has surpassed 500 enterprise clients, including Fortune 500 companies, and its Annual Recurring Revenue (ARR) has exceeded 50 million RMB.
AI's trajectory is shifting from raw parameter counts to practical application, with "memory" emerging as a critical differentiator. Hongxiong AI's "memory science" approach, integrating a dedicated memory system with large language models, addresses systemic issues like context loss and high operational costs inherent in current architectures. This focus on a "hippocampus" for AI, mimicking biological memory functions, represents a potential paradigm shift towards more robust and efficient AI systems capable of sustained, context-aware interactions. The company's expansion into consumer applications suggests a belief in the broad applicability of this memory-centric AI, which could redefine user experiences across various digital platforms. As the AI landscape matures, companies that can demonstrably solve core operational challenges and offer more intuitive, persistent AI interactions are likely to gain significant market advantage.
AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.