Yunzhisheng Unveils ShanhAI Large Model, Marking a Major Leap Toward AGI

Riding the Clouds, Breaking the Waves, and Journeying to the Mountains and Seas. On May 24, the launch event for Unisound’s Shanhai Large Language Model and its achievements was held in Beijing, featuring live demonstrations of the model’s ten core capabilities and the release of a series of product applications tailored to the needs of various industries.

 Unveiling of the Shanhai Large Model by Yunzhisheng

Huang Wei, Founder and CEO of Unisound, stated that the release of the Shanhai Large Model marks a significant milestone in the upgrade of Unisound’s AGI technology architecture. Building on the Shanhai Large Model, Unisound will develop AI 2.0 solutions based on the Model-as-a-Service (MaaS) model. By enhancing industry-specific capabilities in areas such as the Internet of Things (IoT) and healthcare atop its general-purpose functionalities, Unisound aims to provide customers with more intelligent and flexible solutions, thereby unlocking greater commercial potential for the industrial application of AI technologies. Meanwhile, Unisound will continue to upgrade the capabilities of the Shanhai Large Model, with the goal of achieving general-purpose performance comparable to ChatGPT within the year, and surpassing GPT-4 in multiple vertical domains, including healthcare, IoT, and education.

Speech by Huang Wei, Founder and CEO of Unisound

At the press conference, Unisound delivered explanations and demonstrations via voice input and real-time interaction, focusing on the ten core capabilities of the Shanhai Large Model. These include seven general-purpose capabilities—language generation, language understanding, knowledge-based question answering, logical reasoning, coding proficiency, mathematical ability, and safety and compliance—and three industry implementation capabilities: plugin extension, domain enhancement, and enterprise customization.

 Live Demonstration of the Core Capabilities of the Shanhai Large Model

As one of the most representative capabilities of generative AI, language generation is a foundational capability of the Shanhai Large Model. The Shanhai Large Model can not only generate fluent and coherent texts across various formats—including news articles, essays, novels, emails, classical Chinese poetry, and couplets—but also support diverse language generation tasks such as content creation, summarization, and translation through multilingual, multi-genre, and multi-style approaches. Furthermore, it supports controllable text generation under multiple constraints.

In terms of language understanding, the Shanhai large model is capable of comprehending complex ideas by integrating context, common sense, and knowledge to deeply grasp the true meaning of sentences and the emotions they convey.

In the realm of knowledge-based question answering, the Shanhai Large Language Model demonstrates robust performance in terms of both the depth and breadth of knowledge, as well as its capability for interdisciplinary knowledge integration.

Logical reasoning, mathematics, and coding capabilities collectively embody the chain-of-thought prowess underlying the Shanhai Large Model. Currently, the Shanhai Large Model possesses the ability to observe, compare, analyze, synthesize, abstract, generalize, judge, and reason about various phenomena, while accurately and coherently articulating its thought processes; its mathematical and coding capabilities are also undergoing continuous iterative evolution.

Furthermore, the safety and compliance capabilities of the Shanhai Large Model ensure that its outputs are lawful and compliant, enable positive guidance, and mitigate potential security risks.

Large language models serve merely as the foundation for knowledge and capabilities. The plugin extensions, domain-specific enhancements, and enterprise customization features of the Shanhai Large Model further expand its capability boundaries, better meeting the growing demands across various industries for greater flexibility, versatility, and practicality in large language models.

Amid the sweeping wave of AGI, the deep integration of large models with specific application scenarios has become inevitable. To enhance the practical deployment of large models in real-world settings, Unisound has continued its consistent U+X strategy, which leverages U (AI technology and product capabilities) to deeply integrate with X (industry-specific application scenarios), thereby addressing profound industry challenges.

At the event, Unisound unveiled a range of industry-specific applications built on its Shanhai large model, tailored to diverse scenario needs. These solutions are accelerating the intelligent transformation across countless industries by enhancing efficiency, reducing costs, and improving user experience.

 Multiple Product Applications Based on the Shanhai Large Model Were Released On-Site

In the healthcare sector, Unisound, with its years of deep expertise in the industry, has comprehensively upgraded the intelligence level of its entire medical product portfolio by leveraging its accumulated data and experience alongside the Shanhai Large Language Model. It has launched three major medical applications: an Operative Note Writing Assistant, an Outpatient Medical Record Generation System, and an Intelligent Commercial Insurance Claims Processing System, marking a transition from assistant-level support to expert-level capability.

For sales scenarios, Unisound has upgraded its Yunbei Sales Management System by integrating the Shanhai Large Model. Through a four-step approach—customer profiling, intelligent review of sales calls, high-quality sales dashboards, and automated generation of follow-up tasks—the system keenly identifies strengths and areas for improvement in the sales process, gains more accurate insights into customer needs, and enables efficient customer acquisition.

In the context of knowledge management, traditional enterprise internal knowledge management processes often present employees with challenges such as numerous answer versions, information overload and complexity, and imprecise retrieval when searching for solutions. To address these pain points, Unisound has upgraded its existing knowledge management service system by leveraging the Shanhai Large Language Model to create an enterprise-grade "New Bing." This solution provides concise answers to help employees better understand internal technical and professional documents, while offering precise source tracing for every response.

In educational settings, traditional English learning often lacks a communicative environment and error-correction mechanisms, leaving many learners with poor listening and speaking skills despite years of strenuous study. Leveraging the Shanhai Large Language Model, Unisound helps English learners improve their spoken proficiency through a three-tiered correction system encompassing pronunciation guidance, grammar correction, and dialogue generation, thereby putting an end to "mute English."

Targeting smart IoT scenarios, Unisound will comprehensively upgrade its core smart IoT products by deeply integrating them with the Shanhai Large Language Model, creating a true personal assistant. This advancement will elevate interactions from simple command-based exchanges to human-like conversations, seamlessly connecting the IoT ecosystem and services.

On the day of the press conference, Unisound entered into strategic partnerships with China Construction Electronics, JD Technology, and 360, engaging in deep collaboration with these partners to drive the deployment and application of the Shanhai Large Model across various sectors, jointly embracing the wave of the AGI era.

 Unisound and China Construction Electronics Sign Strategic Cooperation Agreement

 Unisound and JD Technology Sign Strategic Cooperation Agreement

 Unisound and 360 Sign Strategic Cooperation Agreement

From entering the deep learning arena in 2012, to later building full-stack AI capabilities, and then competing in the large language model race, Unisound has undergone successive rounds of market trials and self-transformation over the past decade, evolving from an emerging AI voice player into a leader in the artificial intelligence sector. Behind its ability to navigate market cycles lies a reflection of what has changed and what has remained constant in Unisound’s corporate DNA.

What changes is Unisound’s continuous self-reinvention in keeping pace with the times. What remains unchanged is its strategic commitment to investing in technology.

As China’s AI industry collectively pivoted toward investing in computer vision, Unisound remained steadfast in its belief that language is not merely a tool for communication but also a carrier of thought and knowledge—the crown jewel of AI. While many questioned why a four-year-old company would insist on building its own supercomputing center, Unisound persisted, dedicating nearly a year to constructing the Atlas platform. It subsequently rolled out a comprehensive layout of full-stack AI technologies, including knowledge graphs and multimodal systems, thereby evolving from a voice interaction company into deeper domains and achieving a technological upgrade from “sound” (perception) to “knowledge” (cognition).

Unisound remains steadfast in this commitment. In late 2022, as ChatGPT gained widespread attention beyond its initial circle, many teams were debating whether and how to keep pace. Unisound, however, recognized that the long-anticipated era of Artificial General Intelligence (AGI) envisioned by its U+X strategy had arrived, and that all its prior accumulations were poised for a breakthrough. The company swiftly mobilized its R&D teams to expand computing power, tackle engineering optimization, and refine data selection, building upon its Atlas Intelligent Computing Platform and DCML Model Factory. Within just a few months, it completed tasks including computing capacity expansion, algorithm validation, parallel acceleration, and data optimization, achieving a core architecture upgrade centered on GPT. This culminated in the successful launch of the Shanhai Large Language Model, marking a new journey toward AGI.

Unisound AI’s Triple Leap

It is reported that the AGI technology upgrade represented by the Shanhai Large Model marks Unisound’s third major technological leap since its founding 11 years ago. Unisound appears to consistently anticipate trends, seizing the initiative in technological advancements with keen insight. Amidst the ever-changing landscape of the industry, Unisound recognizes that only by embracing the tide of change and steadfastly adhering to its commitment to technological innovation can it stand out through successive rounds of intense market consolidation.

The coming decade will be a dynamic new era, belonging to the age of Artificial General Intelligence (AGI) and marking a period of tremendous advancement in productivity. In its future exploration and development, Unisound will continue to uphold the spirit of innovation, openness, and collaboration, working hand in hand with partners to advance artificial intelligence technology and bring new growth paradigms and endless possibilities to industries across the board.