Deploy SAM
Enterprise Cost Transparency

Pay your cloud provider.
See every token.

No opaque bundles. Compute costs go to your cloud provider at your negotiated rate. LLM costs go through your API key agreements. SAM gives you dashboards to track it all.

Direct Cloud Pricing

VMs are provisioned in your cloud account. You pay Hetzner, Scaleway, Vultr, Infomaniak, DigitalOcean, UpCloud, or GCP directly. No markup, no middleman. Your existing volume discounts and enterprise agreements apply.

The SAM control plane runs on Cloudflare Workers. Self-hosters need the Workers Paid plan for Durable Objects and Cloudflare Containers, while usage-based costs stay low for small to medium teams.

Token Usage, by Model and by Day

Settings → Usage shows each user their own AI usage broken down by model and by day, tracked through the SAM AI proxy and Cloudflare AI Gateway. Only traffic routed through the platform proxy is metered this way — bring-your-own-key traffic that bypasses the proxy is not counted here.

Settings → Usage page showing token usage and cost per model, a per-day usage chart, and progress against a monthly budgetSettings → Usage page showing token usage and cost per model, a per-day usage chart, and progress against a monthly budget Token usage by model and by day, against a monthly budget

Budget and Quota Controls

Set daily token limits and a monthly cost cap per user. Compute has its own quota, tracked in node-hours against a monthly limit — BYOC users who connect their own cloud credentials are exempt, since that spend never touches SAM's billing.

Settings → Usage compute section showing the monthly quota with hours remaining, node-hours, vCPU-hours, platform vCPU, active nodes, and the active node list with observed and configured hardwareSettings → Usage compute section showing the monthly quota with hours remaining, node-hours, vCPU-hours, platform vCPU, active nodes, and the active node list with observed and configured hardware Compute usage in node-hours and vCPU-hours against the monthly quota
Control Scope Enforcement
Daily token limit Per user Input/output token ceiling, resets daily
Monthly cost cap Per user USD ceiling on SAM-proxied AI spend
Compute quota Per user Node-hours per month; exempt for BYOC users
Admin allowance ceiling Per user (admin-set) Max limit a user can self-assign

Admin Cost Dashboard

Admin → Costs aggregates LLM spend by model, by day, and by user, with a projection, alongside compute cost. Admin → Usage breaks out per-user node usage across the installation. Superadmins get one place to see what the whole team is spending without parsing cloud provider invoices.

Admin → Costs dashboard showing LLM spend by model, day, and user, a cost projection, and compute costAdmin → Costs dashboard showing LLM spend by model, day, and user, a cost projection, and compute cost LLM spend by model, day, and user, with a projection, plus compute cost
Admin → Usage node usage table listing each user with node-hours, vCPU-hours, platform hours, and active node count for the billing periodAdmin → Usage node usage table listing each user with node-hours, vCPU-hours, platform hours, and active node count for the billing period Per-user node usage across the installation for the billing period
Open Source Scale

Get your coding agents off your laptop

SAM is open source. Deploy it to your Cloudflare account and connect your agent. Add a cloud provider when you want full environments on your own servers.

Deploy SAM Star on GitHub