Skip to content

Documentation

CoreValley Documentation

Guides and reference for running AI workloads on CoreValley, Nepal's sovereign GPU cloud. Everything here describes the platform as it ships; sections that are still being written say so.

Early access — documentation in progress

Early access is open to enterprises on NVIDIA H200. More GPUs, and access for more teams, are coming soon. These pages are being written alongside the platform, and each section below describes what it will cover. For anything you need now, email info@corevalley.ai — an engineer in Kathmandu answers within one working day.


Start here#


How the documentation is organised#

SectionWhat it covers
PlatformThe four products — GPU pods, JupyterHub, Model endpoints and Dedicated servers — and how they share projects, storage and keys. They were previously called GPU Workspace, AI Lab and Inference API.
HardwareWhich NVIDIA GPUs are available (H200 today; RTX PRO 6000 Blackwell and L40S coming soon), slice profiles, memory, bandwidth and networking.
BillingHow usage is metered, how invoices are produced and how to pay in NPR.
GuidesTask-oriented walkthroughs, starting with the quickstart.
SupportService levels, maintenance windows and how to reach the team.

Conventions used throughout: commands are shown for the corevalley CLI and the console side by side where both exist; times are Nepal Time (UTC+05:45) unless marked UTC; the region — our hydro-powered datacenter in Kathmandu — is referred to as np-ktm-1.

These docs cover the platform as it ships today. For anything not documented here, email info@corevalley.ai.