范式智能近日宣布，"信创模盒"ModelHub XC已完成108个主流AI模型在摩尔线程GPU上的适配认证， 涵盖文本生成、视觉理解、多模态问答等多种任务类型，预计未来半年内将扩展至千量级，为国产算力 生态注入持续动能。 值得一提的是，在本次批量适配过程中，作为年内登陆科创板的国产GPU企业，摩尔线程的硬件在量化 模型方面展现出显著优势。其GPU凭借对低精度数据类型的硬件级支持、优化的指令集与缓存机制，有 效降低了模型显存占用并提升推理速度，使量化模型在实际部署中兼具效率与能效。通过精细化校准与 优化，适配模型在性能提升的同时，也确保了推理精度满足商业落地要求。据报道，摩尔线程在11月24 日正式启动科创板发行，发行价为114.28元/股，创下2025年以来A股新股发行价新高。 在AI推理效率成为产业落地核心挑战的今天，如何实现模型在国产芯片上的高效、稳定运行，已成为 推动算力生态走向成熟的关键。作为应对，范式智能依托自研的EngineX引擎技术，重点突破模型在国 产芯片上的兼容性与运行效率，显著降低开发者的部署门槛。 目前，"信创模盒"ModelHub XC已完成包括Mata、千问、Deepseek、混元、 ...

Core Insights - Paradigm Intelligence announced that its "ModelHub XC" has achieved over 1,000 certified models in just two months, four months ahead of schedule, marking significant progress in the domestic AI ecosystem [1][12] - The platform supports a diverse range of models, from general large language models to specialized vertical models and cutting-edge innovations, providing a solid foundation for the coordinated development of domestic AI hardware and software [1][12] Group 1: Platform Development - The "ModelHub XC" platform was officially launched on September 22, 2025, aiming to address the compatibility issues between deployed models and underlying chip architectures [2] - Key milestones include the successful adaptation of complex vertical models on domestic chips, achieving commercial-grade performance standards [4] - The platform has demonstrated strong ecological expansion capabilities by completing the adaptation of 108 models in a single batch, covering various task types [11] Group 2: Technical Innovations - The platform has achieved significant breakthroughs in adapting advanced models, such as the DeepSeek-OCR, which utilizes visual modality to compress text information, addressing efficiency challenges in large language models [6] - The MiniMax-M2 model, a leading open-source agent model, has been adapted for domestic chips, showcasing its global competitiveness with 230 billion parameters [8][9] Group 3: Future Outlook - The platform aims to accelerate towards a "ten thousand model" ecosystem within a year, continuously expanding model scale and chip support [13] - The focus will be on maintaining a rapid update pace to build a more complete and efficient domestic AI infrastructure [13]

Core Insights - The launch of "ModelHub XC" by Paradigm Intelligence has achieved over 1,000 model adaptations within two months, four months ahead of schedule, marking significant progress in the domestic AI ecosystem [1][11] - The platform supports a diverse range of models, including general large language models, specialized vertical models, and cutting-edge innovative models, providing a solid foundation for the coordinated development of domestic AI software and hardware [1][12] Development Timeline - **Launch Date**: The platform was officially launched on September 22, 2025, addressing the compatibility issues between deployed models and underlying chip architectures, which has been a barrier to the large-scale implementation of AI [2][12] - **Vertical Model Adaptation**: On October 17, 2025, the platform completed the adaptation and deep optimization of the wind tunnel calculation model on the domestic chip Xiwang S2, achieving commercial-grade performance [4] - **Frontier Model Adaptation**: On November 1, 2025, the innovative model DeepSeek-OCR was successfully adapted for testing on various domestic computing cards, showcasing significant technical innovation [6] - **Agent Model Deployment**: On November 17, 2025, the efficient agent model MiniMax-M2 was adapted for deployment on the Ascend 910B4 chip, demonstrating global leadership in model capabilities [7] - **Batch Adaptation Achievement**: On November 25, 2025, the platform achieved large-scale adaptation of 108 models on the Moer Thread chip, highlighting its strong ecological expansion capabilities [9] Platform Capabilities - The platform is driven by the EngineX engine, enabling "plug-and-play" deployment of models on domestic chips, significantly shortening deployment cycles and resolving compatibility issues [12] - The model ecosystem is rich and diverse, covering a wide range of models and supporting major domestic computing platforms [12] - The platform offers professional services for model adaptation, backed by a team of hundreds of engineers to ensure successful adaptation and stable operation in domestic environments [12] Future Outlook - The platform aims to accelerate towards a "ten thousand model" ecosystem within a year, continuing to expand model scale and chip support [14] - The company plans to maintain a rapid update pace to build a more complete and efficient domestic AI infrastructure [14]

Core Insights - Paradigm Intelligence announced that its "ModelHub XC" has completed the adaptation certification of 108 mainstream AI models on Moore Threads' GPU, covering various task types such as text generation, visual understanding, and multimodal Q&A, with plans to expand to a thousand models in the next six months [1][3] - Moore Threads, a domestic GPU company that recently launched on the Sci-Tech Innovation Board, demonstrated significant advantages in quantized models during the adaptation process, showcasing hardware-level support for low-precision data types, optimized instruction sets, and cache mechanisms [1] - The launch price of Moore Threads' shares was set at 114.28 yuan per share, marking a new high for A-share IPO prices since 2025 [1] Company and Industry Summary - The "ModelHub XC" serves as an AI model and tool platform aimed at the domestic computing power ecosystem, providing a comprehensive solution from model training and inference to deployment, while also fostering community and service functionalities [4][5] - The EngineX engine technology developed by Paradigm Intelligence focuses on improving model compatibility and operational efficiency on domestic chips, significantly lowering deployment barriers for developers [1][3] - The adaptation process included models from various series such as Mata, Qianwen, Deepseek, and others, effectively addressing the bottlenecks in model compatibility and scalability on domestic chips [3]

Core Insights - Paradigm Intelligence recently announced that its "ModelHub XC" has completed the adaptation certification of 108 mainstream AI models on Moore Threads GPUs, covering various task types such as text generation, visual understanding, and multimodal Q&A, with plans to expand to a thousand models in the next six months, injecting continuous momentum into the domestic computing power ecosystem [1][3] - Moore Threads, a domestic GPU company set to launch on the Sci-Tech Innovation Board, has demonstrated significant advantages in quantized models during this adaptation process, with its GPUs effectively reducing model memory usage and enhancing inference speed through hardware-level support for low-precision data types and optimized instruction sets [1] - The official launch of Moore Threads on the Sci-Tech Innovation Board is scheduled for November 24, with an issuance price of 114.28 yuan per share, marking a new high for A-share IPO prices since 2025 [1] - The efficient and stable operation of models on domestic chips is a key challenge for the industry, and Paradigm Intelligence is addressing this by leveraging its self-developed EngineX engine technology to improve model compatibility and operational efficiency on domestic chips, significantly lowering deployment barriers for developers [1][5] Summary by Sections ModelHub XC Overview - ModelHub XC is an AI model and tool platform aimed at the domestic computing power ecosystem, providing a comprehensive solution that covers the entire process from model training and inference to deployment, while also serving community and service functions [5] EngineX Engine - The EngineX engine serves as the underlying support system for ModelHub XC, enabling "engine-driven, multi-model plug-and-play" capabilities, effectively addressing the bottlenecks in model compatibility and scale support on domestic chips [3][5]