공지 • May 05
Lobo Technologies Ltd Completes New Round Of Upgrades With Deepseek V4 Model Integration
LOBO TECHNOLOGIES LTD. announced that its independently developed Claw AI Agent platform has completed a new round of upgrades and officially integrated the DeepSeek V4 large model. The Platform now supports collaborative scheduling of three mainstream AI models: OpenAI, Google Gemini, and DeepSeek, further strengthening the Company’s product competitiveness in AI agent orchestration, complex task automation, and enterprise-grade AI implementation. The Platform features a built-in intelligent routing layer that automatically matches the most suitable underlying model for each call based on multi-dimensional parameters including task type, context length, response latency, and cost budget. DeepSeek V4, integrated to the Platform through this upgrade, is one of the open-source foundation models released in April 2026. Its Pro version adopts a 1.6-trillion-parameter Mixture-of-Experts (MoE) architecture, activating approximately 49 billion parameters per forward pass, and supports an ultra-long context window of 1 million tokens. Public benchmark results demonstrate outstanding performance in coding (LiveCodeBench 93.5%), mathematical reasoning, and long-text comprehension. For LOBO customers, this capability enhancement means: Long-document scenarios: processing hundreds of pages of contracts, technical specifications, or market analysis reports in a single session. Complex decision-making: maintaining decision quality in multi-variable trade-off and long-chain reasoning tasks. High-concurrency response: leveraging inference efficiency advantages of the MoE architecture to improve concurrency and reduce latency. The upgraded Platform supports a “Thinking Mode,” enabling users to view the AI agent’s reasoning process in complex tasks such as product planning, customer benchmarking, country-specific strategies, customer acquisition, and supply chain assessment. By providing transparent reasoning chains, the Platform makes enterprise-level decision support more transparent and auditable, meeting compliance and internal control requirements, including those applicable to listed companies. For complex business requirements beyond the scope of a single conversation, the Platform automatically completes the full process of “goal recognition ? task decomposition ? multi-step workflow planning.” After user confirmation, sub-agents proceed step-by-step in a workstation environment. The entire execution supports real-time progress visualization, cross-step data flow, and structured delivery of documents and tasks. For high-risk actions involving external publishing, fund transfers, and large-scale data modifications, the Platform features a built-in human-machine collaboration approval mechanism. The AI agent automatically triggers a user approval panel before executing such actions, allowing users to approve or reject actions within a set timeframe, balancing AI automation and enterprise risk control. The Platform has built a professional AI assistant matrix covering core positions such as product planning, technical evaluation, marketing and operations, foreign trade sales, and customer service. Each intelligent assistant is equipped with targeted domain knowledge, tool invocation capabilities, and scenario-based workflows, providing professional support in manufacturing product strategy, Bill of Materials (BOM) construction and evaluation, foreign trade contract review, cross-border customer communication, and other use cases. The Platform provides comprehensive role-based access control (RBAC) that allows four-tier permissions, as well as a token billing system, complete audit logs, context memory management, and multi-modal input capabilities. All AI calls, tool executions, and document modifications are traceable, helping enterprises, including listed companies, to meet compliance audit and internal risk control requirements. The Platform has realized a complete “Dialogue ? Planning ? Execution ? Delivery” workflow closed loop. After users submit business requirements via natural language dialogue, the AI agent automatically completes intent understanding, goal recognition, task decomposition, workflow orchestration, and final deliverable output, generating structured documents and executable tasks. This closed loop has undergone internal verification across multiple real-world scenarios, including manufacturing product strategy, BOM creation and evaluation, foreign trade contract review, and cross-border market analysis, and is now in daily use by the Company’s internal business teams.