Full-path metric coverage
Covers GPU compute, host resources, host networks, inter-server networks, and training/inference logs.
Products and services
XOPS｜AI computing operations system
For large-scale training and inference on AI clusters, build an integrated operations platform across compute, network, and storage for faster problem localization and efficient troubleshooting.
Break information silosCentered on AI AgentsReduce manual effort
Product positioning
Turn full-path observability into an AI Agent operations loop
XOPS unifies GPU, host, network, and training/inference job monitoring, then uses visualization and multi-agent collaboration to connect anomaly detection, root-cause localization, ticket analysis, and closed-loop handling.
Core capabilities
- GPU compute and host resource monitoring
- Host network and inter-server network monitoring
- Unified view of training/inference logs and job status
- AI Agent root-cause localization and ticket closure
Core capability architecture
From unified observability to intelligent handling
For large-scale training and inference, metrics, status, and logs are presented together to help locate issues quickly.
Visual monitoring dashboards
Present compute, network, and job metrics together to reduce silos and switching across systems.
Multi-agent collaborative automation
Jointly analyze alerts, metrics, logs, and tickets to assist root-cause localization, automatic ticket replies, and closed-loop handling.
Network operations Agent Benchmark
Evaluate large models on long-horizon network operations tasks, build an operations experience library and Skills, and recommend the best-performing model.
Value and results
Operations value
Unified data and intelligent analysis improve completeness of information, response speed, and automation in AI cluster operations.
Complete information
Break silos among compute, network, and job data with a unified view.
Faster response
Locate root causes quickly and shorten MTTR.
AI intelligence
Agents assist independent diagnosis, automatic ticket replies, and end-to-end closed-loop handling, reducing manual effort.
Products and services
Other products and services
Networking, tuning, operations, and professional services work together to cover the full lifecycle of AI cluster networking.
AI cluster networking
ZCubeNext-generation AI cluster networking architecture
Replace the traditional multi-tier tree with a fully flat interconnect, remove Spine-layer switches, and combine single-rail plus multi-rail hybrid access with dedicated routing so GPU communication inside the cluster takes shorter, more balanced paths.
Learn about the productAI computing tuning
YCOSAI computing tuning system
For large-scale AI workloads, systematically tune training/inference frameworks, communication libraries, network configuration, and host configuration.
Learn about the productProfessional services
Full Lifecycle ServicesComplete delivery from planning to go-live
One-stop delivery of network devices, compute-network tuning, and operations services to keep AI cluster networks stable, communication efficient, and SLAs met, helping clusters go live faster.
Learn about the productStart with the network
Make the next AI data center more efficient, starting with the network
Whether you are building a new cluster, upgrading an existing network, tuning performance, or deploying intelligent operations, we can assess how the network can unlock more effective compute.