ModelHub XC Surpasses 2000+ Model Adaptations, Enhancing Key Capabilities Like Inference and Code Support

FOURTH PARADIGM (06682) recently announced a milestone update for its ModelHub XC platform, which has now certified over 2,000 adapted AI models—surpassing its target ahead of schedule by half a month. Following the achievement of "1,000-model adaptation" in November, the platform has accelerated upgrades to its automated computing engine, improving both the speed and quality of model adaptation. This progress further strengthens China's domestic AI ecosystem in terms of model coverage breadth and reliability.

The newly adapted models were contributed by 100+ open-source organizations and developer communities, covering mainstream frameworks such as Qwen, Llama, Mistral, Phi, and DeepSeek. Key enhancements include expanded support for inference and code-related capabilities. The platform primarily supports Ascend 910B3/910B4 chips while extending compatibility to domestic computing platforms like MetaX Cloud C500, Hygon K100AI, and Iluvatar CoreX 100.

Beyond quantity, the 2,000+ certified models emphasize deployment certainty. ModelHub XC employs a tagged matching system (model-hardware-software) to minimize trial-and-error for developers. For enterprises with proprietary or complex industry models, the platform offers value-added adaptation services to optimize "last-mile" tuning, reducing redundant adaptation costs and deployment barriers. This accelerates AI adoption in critical sectors such as infrastructure, public security, transportation, and energy.

As large-scale AI models rapidly evolve, aligning domestic computing power with these models has become pivotal for China's self-sufficient AI industry. ModelHub XC’s accelerated adaptation marks a shift from basic operability to efficient performance on local hardware.

FOURTH PARADIGM invites developers, enterprises, and industry partners to join the ModelHub XC ecosystem, collaborating on model adaptation, technical optimization, and scenario-based solutions to deepen the integration of domestic computing power and AI applications, driving intelligent transformation across industries.

About ModelHub XC: A dual-purpose platform (community + service) for China’s computing ecosystem, ModelHub XC facilitates AI innovation and deployment on domestic hardware, providing end-to-end solutions from model training and inference to deployment.