Deploy a qualified GPU workload, verify acceptance and add infrastructure or AI Production Care with a written responsibility boundary.
Start with deployment, add infrastructure care, or extend the scope to a named production AI workload. GPU capacity remains a separate line.
Build the agreed runtime, deploy the workload and close the project against written acceptance checks.
Maintain the quoted GPU node and runtime boundary after deployment.
Add a named operating boundary around a production serving or AI workload.
The customer keeps model, data and application ownership unless the service plan adds a specific responsibility. ZenoCloud owns only the listed infrastructure and operating actions.
The operating scope starts only after the workload, environment and acceptance path are understood.
Record model, framework, precision, concurrency, memory, data, network, observability and acceptance targets.
Install the agreed baseline, preserve the customer environment and run the written verification checks.
Monitor, maintain, respond, record evidence and hand application or model work to the correct owner.
The quote states the runtime, drivers, container environment, serving or batch framework, access, observability and workload checks included. Anything not written into the acceptance scope remains excluded.
No. GPU Infrastructure Care covers the agreed host, runtime and infrastructure boundary. AI Production Care adds a defined serving or workload boundary, such as deployment health, model endpoint signals and coordinated incident ownership.
Yes where the platform is supported, access is available and provider responsibilities are documented. ZenoCloud does not require an account transfer by default.
Not by default. Model selection, training, fine-tuning and application behavior stay with the named model or application owner unless bought as a separate service.
No automatic unpaid evaluation is promised. A validation or benchmark run is available only when qualified and stated in the written quote.