en.Wedoany.com Reported - WorldClaw and Baidu Intelligent Cloud announced a strategic partnership on July 3, focusing on collaboration around AI large models and the Agent ecosystem, with initial implementation in model service integration. The Baidu ERNIE 5.0 series of large models has been listed on WorldRouter, the model aggregation platform under WorldClaw, via the Baidu Intelligent Cloud Qianfan Large Model Platform, making it accessible for developers to call upon.
WorldClaw is an AI infrastructure project supported by World Liberty Financial (WLFI), positioned as an open operating system for the Agent economy. The platform provides a unified model invocation entry point through WorldRouter, aggregating over 300 large language, image, and video models. Developers can call upon the capabilities of multiple models through a single API without needing to interface with different vendors separately. WorldClaw has also launched the AgentPay SDK, enabling AI agents to execute tasks, make payments, and settle under compliance policies and manual approval mechanisms.
In this collaboration, Baidu Intelligent Cloud has integrated the ERNIE 5.0 series of large models into WorldRouter via the Qianfan Large Model Platform and completed protocol adaptation at the gateway layer. The ERNIE 5.0 series employs native full-modal unified modeling technology and a ultra-sparse Mixture of Experts (MoE) architecture. The flagship model has over 2.4 trillion parameters, supporting the understanding and generation of multiple modalities including text, images, audio, and video. After listing, developers can directly invoke the ERNIE model series within WorldRouter's existing interface framework, with model invocation, result return, tool invocation, and usage statistics all operating under the same set of interfaces.
Beyond model service integration, the two parties will further collaborate on Skill co-creation, A2A scenario exploration, and AgentOS development. Baidu Intelligent Cloud plans to package capabilities such as Vibe Coding, video generation, and model security governance into Skills and list them on WorldClaw's Skill marketplace. Additionally, both parties will explore the application of capabilities like task orchestration, multi-agent workflows, and long-chain task management within AgentOS.
Currently, large model applications are evolving from single model invocation to multi-model combinations. Different tasks impose differentiated requirements on model capabilities, response speed, and invocation cost. A unified model entry point is becoming key infrastructure connecting model services with Agent application development, helping model capabilities enter developer workflows more quickly.
