Aurora Mobile Rolls Out 744B-Parameter GLM-5, Cuts AI Hallucination by 35 Points
Aurora Mobile integrated the 744 billion-parameter GLM-5 model into its GPTBots.ai platform, achieving a record −1 AA-Omniscience score and 35-point improvement in reliability over its predecessor. The open-source, enterprise-grade AI offers automated document generation and native agentic capabilities, boosting productivity and cost-effectiveness for clients.
1. GLM-5 Integration and Technical Specs
GPTBots.ai now supports the open-source GLM-5 model with 744 billion parameters and MIT license weights. This integration provides enterprises full access to advanced Mixture-of-Experts and DeepSeek Sparse Attention technologies in a no-code platform.
2. Reliability and Automation Improvements
GLM-5 achieved a record −1 AA-Omniscience Index score and reduced hallucination rates by 35 points over its predecessor. Native Agent Mode features autonomously generate professional .docx, .pdf, and .xlsx outputs, streamlining report creation and workflow automation.
3. Cost Efficiency and Enterprise Readiness
The platform’s scalable architecture and competitive pricing lower entry barriers for large-scale AI deployment. Open weights and no vendor lock-in empower organizations to customize and deploy frontier-level AI without significant infrastructure investment.
4. Strategic Impact for Aurora Mobile
By being first to integrate GLM-5, Aurora Mobile strengthens its position in the enterprise AI market and enhances its competitive edge. Improved performance, reliability, and cost-efficiency could drive new client acquisitions and accelerate revenue growth.