Baidu Intelligent Cloud Unveils Bage AI Computing Platform 5.0

At the 2025 Baidu Cloud Intelligence Conference, Shen Dou, Executive Vice President of Baidu Group and President of Baidu Intelligent Cloud Business Group, announced the official upgrade of the Bage AI Computing Platform to version 5.0. This release focuses on breaking efficiency bottlenecks in AI computing through enhancements in four key areas: networking, computing power, inference systems, and training-inference integration.

Networking Improvements

The fifth version of the Bage platform achieves faster communication speeds and lower latency, significantly boosting the efficiency of model training and inference. These improvements are critical for large-scale AI applications requiring real-time data processing.

Enhanced Computing Power

Following the release of the Kunlun Chip Super Node at the Create 2025 Baidu AI Developer Conference, the upgraded Bage platform now integrates this technology into Baidu Intelligent Cloud's public cloud service. This provides users with supercomputing capabilities, enabling them to run trillion-parameter models with just a few minutes and a single cloud instance.

Inference System Upgrades

The new version introduces three core strategies—decoupling, adaptive, and intelligent scheduling—to improve throughput and reduce latency. These advancements ensure smoother performance for complex AI workloads.

Training-Inference Integration

Baidu also unveiled the Bage Reinforcement Learning Framework, designed to maximize computing resource utilization. This framework enhances both training and inference efficiency, making it easier for developers to deploy large-scale AI models.

Key Points:

• Faster networking: Reduced latency and improved communication speeds.
• Kunlun Chip Super Node: Enables trillion-parameter model execution.
• Inference optimizations: Decoupling, adaptive strategies, and intelligent scheduling.
• Reinforcement Learning Framework: Boosts training-inference integration efficiency.