ARVICA / DOCUMENTATION

Arvica overview

Choose compute, prepare your deployment and manage the service through one workspace.

Your compute workspace

Arvica is the customer-facing AI compute platform and service layer. Browse GPU specifications, compare configurations and submit a deployment request. Your workspace brings together requests, quotes, orders, recorded usage and support.

From request to access

The live catalogue exposes supported offers; accepting a current quote and confirming Deploy can provision a billable single-node instance. Dedicated-node and enterprise cluster enquiries follow a separate public RFQ: capacity, topology, location, start date and terms are agreed before activation. A catalogue specification or RFQ alone does not reserve a machine.

Service activation

For live on-demand compute, choose a supported offer and review the quote and billing terms before Deploy. For dedicated or reserved capacity, submit a free public enquiry and agree the deployment proposal first. The workspace manages activated services, usage and support.

Browse GPUs →

Create your account →Start with an email address and complete verification.Your first deployment request →Prepare a complete configuration before submitting.GPU and instance selection →Select memory and topology for the workload you will run.Deployment location →Confirm the actual location with the selected configuration.SSH and remote development →Connect only after instance access details are confirmed.Networking and access →Plan application ports and access controls with your instance.Images and GPU environments →Make the software stack part of your deployment requirements.Training and fine-tuning →Plan a reproducible training run and recovery path.Inference and image generation →Size the service around latency, throughput and memory.Storage and data retention →Understand what is requested, recorded and retained.Billing and credit →Read the confirmed contract and its transaction history.Cost planning and reservations →Compare total service cost rather than a GPU headline rate.Teams and account security →Keep access aligned with each person’s role.Customer API quick start →Use scoped keys against the environment where they were issued.Troubleshooting and support →Send enough context to make the issue reproducible.GPU environment setup & SSH →Prepare and check a Linux GPU host before installing a model server.Run a vLLM inference server →A small-model smoke test with local-only access, health checks and cleanup.Run Ollama on a GPU →Start a small model, verify the API and keep model data in a persistent Docker volume.Capacity enquiry to your first connection →This guide covers dedicated-node and enterprise capacity enquiries. Live on-demand deployment has a separate quote-and-Deploy flow that can start billing after confirmation.Understand your complete quote →Compare the exact service and total cost, not only a GPU headline rate.Storage, stopping and data retention →Specify the data behaviour you need before choosing capacity.