SUPPORT / DOCUMENTATION
How can we help?
From presales sizing and private deployment to 24/7 operations — one place for the answer, the document, or the person who can help.
PRESALES
Presales support
Get the requirement clear before you choose.
GPU sizing for LLM fine-tuning and inference
Estimate VRAM, card type and cluster size from parameters, context and concurrency.
Open guideBuilding an AI general-education lab
Courses, digital teachers, labs and learning analytics in one plan.
Open guidePrivate enterprise AI: boundaries and options
Engineering decisions for data residency, isolation and local model deployment.
Open guideChoosing an enterprise model gateway
Where routing, quotas and billing governance belong in the stack.
Open guideRunning a POC: apply, test, decide
Set metrics, dataset and timeline, then get dedicated test nodes.
Open guideComparing GPU cards: VRAM, interconnect, power
A side-by-side view of VRAM, precision, interconnect and power.
Open guideNeed a tailored plan?
A senior AI infrastructure architect can size the deployment and sketch a hybrid-cloud plan for your workload.
OPERATIONS
After-sales & operations
How to run it after it arrives.
Response is part of the product.
Public service commitments; the signed service agreement prevails.
GPUCloud: create an instance and connect
Create an instance, choose an image, expose ports and connect.
Open guideCloud desktop & workstation onboarding
Multi-device access, environment setup and rendering scenarios.
Open guideCUDA driver & NVLink troubleshooting
A diagnostic sequence for unreachable nodes, hangs and interconnect latency.
Open guidePrebuilt deep-learning images & Docker import
Which prebuilt image to pick, how to pin deps and import your own.
Open guideHigh-performance storage & multi-node sharing
Choosing shared file vs block storage, mounting across nodes, measuring bandwidth.
Open guideDOCUMENTATION
Documentation
Browse by category, or search across the full library. Product docs live in the documentation center.
Open documentation centerGPU sizing for LLM fine-tuning and inference
Estimate VRAM, card type and cluster size from parameters, context and concurrency.
OpenEducationSmartLearn AiEdu-LabBuilding an AI general-education lab
Courses, digital teachers, labs and learning analytics in one plan.
OpenPrivate AISmartBrain WorkMeshPrivate enterprise AI: boundaries and options
Engineering decisions for data residency, isolation and local model deployment.
OpenModel accessSmartLink OpenAPIChoosing an enterprise model gateway
Where routing, quotas and billing governance belong in the stack.
OpenPOC trialSmartCompute GPUCloudRunning a POC: apply, test, decide
Set metrics, dataset and timeline, then get dedicated test nodes.
OpenGPU comparisonSmartCompute GPUCloudComparing GPU cards: VRAM, interconnect, power
A side-by-side view of VRAM, precision, interconnect and power.
OpenCabinet deliverySmartBrain WorkMeshLiquid-cooled cabinet preparation & acceptance
On-site checklist for power, cooling, floor loading and network.
OpenImplementationSmartBrain WorkMeshDeploying and accepting a private AI platform
A delivery sequence from environment checks to model import, permissions and monitoring.
OpenQuick startSmartLink OpenAPISmartLink OpenAPI: three-step onboarding
Change base_url, create an app and inject a key for the first request.
OpenTool ecosystemMCP MarketMCP tools: access and permissions
Connect external tools to agent workflows and scope what they can access.
OpenConsoleSmartCompute GPUCloudGPUCloud: create an instance and connect
Create an instance, choose an image, expose ports and connect.
OpenCloud desktopAI Cloud HostCloud desktop & workstation onboarding
Multi-device access, environment setup and rendering scenarios.
OpenTroubleshootingDocumentationCUDA driver & NVLink troubleshooting
A diagnostic sequence for unreachable nodes, hangs and interconnect latency.
OpenImagesSmartCompute GPUCloudPrebuilt deep-learning images & Docker import
Which prebuilt image to pick, how to pin deps and import your own.
OpenStorage & networkSmartCompute GPUCloudHigh-performance storage & multi-node sharing
Choosing shared file vs block storage, mounting across nodes, measuring bandwidth.
OpenInsightStarwar Research2026 enterprise AI: from tools to agents
Why budgets are moving from point tools to multi-agent collaboration.
OpenGEOStarwar CloudContent visibility in the AI search era
How enterprise sites get cited when buyers ask AI.
OpenGrowthSmartGrowth GJ6The six-dimension AI growth loop
A loop from content production and publishing to lead detection and review.
OpenInsightSmartLearn AiEdu-LabAI in the classroom: making general education work
Policy drivers, course layering and real training compute.
OpenHeterogeneous computeSmartCompute GPUCloudManaging heterogeneous compute in the domestic-GPU era
Unified scheduling, image portability and per-tenant metering across GPU vendors.
OpenCONTACT
Contact us
Call the on-call engineer, email the team, or bring the issue to us directly.
On-call engineers respond to urgent incidents.
ENTERPRISE EMAIL18958102708@189.cnTechnical support and ticket intake.
Business meetings and visits.
Hand the problem to us.
Include the node ID, a screenshot of the error and reproduction steps; the architect can locate it faster.