TeleAI, the Institute of Artificial Intelligence of China Telecom, announced that its AI Flow framework was recognized by Omdia as a pivotal advancement in transforming telecom infrastructure for AI applications. The framework addresses edge generative AI (GenAI) challenges through a device-edge-cloud architecture, optimizing performance and efficiency.
AI Flow recognized by Omdia for telecom AI transformation.
Device-edge-cloud architecture enhances scalability, low-latency inference.
Open-sourced AI-Flow-Ruyi-7B-Preview model on GitHub.
Enables seamless intelligence flow across LLMs, VLMs, and diffusion models.
Led by Professor Xuelong Li, CTO of China Telecom.
Supports real-time collaboration for emergent AI intelligence.
AI Flow’s architecture integrates end devices, edge servers, and cloud clusters to overcome hardware and network constraints, enabling low-latency AI model inference. “AI Flow facilitates seamless intelligence flow, allowing device-level agents to overcome the limitations of a single device,” notes Omdia’s report, highlighting its ability to connect advanced models like LLMs and VLMs across heterogeneous nodes.
The framework introduces familial models, such as the open-sourced AI-Flow-Ruyi-7B-Preview, designed for multi-scale tasks. These models share intermediate features, enabling efficient responses through an early-exit mechanism, reducing computation needs. “Its core innovation lies in the shared intermediate features across models of varying scales,” said TeleAI, emphasizing collaborative inference across distributed systems.
AI Flow fosters dynamic interactions among AI models, creating emergent intelligence that surpasses individual model capabilities. Omdia’s Chief Analyst Lian Jye Su praised its “sophisticated approaches to facilitate efficient collaboration across device-edge-cloud tiers.” Social media buzz includes AI observer EyeingAI calling it “a grounded, realistic take on where AI could be headed” and influencer Parul Gautam noting its potential to “shape the future of intelligent connectivity.”
Led by Professor Xuelong Li, AI Flow addresses deployment challenges, enhancing scalability and sustainability in real-world AI systems. The open-sourcing of AI-Flow-Ruyi-7B-Preview on GitHub, including APIs and scheduling logic, has sparked community interest, with X posts highlighting its dynamic scaling from 1.6B to 7B parameters. TeleAI aims to drive ubiquitous intelligence, impacting telecom and beyond.
TeleAI, the Institute of Artificial Intelligence of China Telecom, is a pioneering team of AI scientists and enthusiasts, working to create breakthrough AI technologies that could build up the next generation of ubiquitous intelligence and improve people's wellbeing. Under the leadership of Professor Xuelong Li, the CTO and Chief Scientist of China Telecom, TeleAI aims to continuously expand the limits of human cognition and activities, by expediting research on AI governance, AI Flow, Intelligent Optoelectronics (with an emphasis on embodied AI), and AI Agents.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.