Open weight and self-hosted

AWS GPU compute cost tracking

EC2 GPU instances running your own models, pulled out of the AWS bill.

How does Camaze work with AWS GPU compute?

Camaze connects to AWS GPU compute with read-only credentials and pulls spend and usage into the same view as every other place you pay for AI.

What gets collected

What Camaze pulls from AWS GPU compute

GPU instances running your own models appear in the AWS bill as EC2 compute and nothing more. Camaze identifies the instances serving inference, attributes their cost to the endpoints and models running on them, and reports cost per million tokens served alongside utilization, so an idle dedicated endpoint reads as a saving rather than as infrastructure.

  • Compute spend, split out of the wider cloud or platform bill
  • Which models and endpoints that compute is serving
  • Utilization, so an idle dedicated endpoint is visible as an opportunity
  • Cost per million tokens served, comparable to hosted providers

See your AWS GPU compute spend in context.

Book a walkthrough and we will show you AWS GPU compute costs next to everything else you pay for AI.

FAQ

Questions people ask

What access does Camaze need for AWS GPU compute?

Read-only billing and usage credentials. Nothing that can spend money, change a deployment, or read your prompts and completions. Details are on the security page.

Does it backfill historical AWS GPU compute spend?

Yes. Camaze pulls the available history on connection, typically the last twelve months, so you start with a trend rather than an empty chart.

Can we split AWS GPU compute spend by team or project?

Yes. Map keys, projects, workspaces or tags to your own org structure once, and every AWS GPU compute charge is attributed from then on. See cost allocation.

Do we have to use AWS GPU compute exclusively?

No. Camaze is built for the opposite case. AWS GPU compute sits alongside every other provider, tool and self-hosted model in the same view, on the same cost basis. It is equally useful if AWS GPU compute is the only thing you run: see single-provider visibility.

See where your AI money is going.

A 30 minute walkthrough of Camaze on your own numbers. You will leave knowing what you spend and where the savings are.

5 minute setup. Read-only. No engineering time needed.