Envoy AI Gateway - CostGraph Documentation

Envoy AI Gateway follows the OpenTelemetry GenAI conventions, so one metric carries both directions of a call. See the upstream metrics reference.

What CostGraph scrapes


Namespace	`envoy-ai-gateway-system`
Pod selector	`app.kubernetes.io/name=envoy-ai-gateway`
Port and path	`1064` `/metrics`
Metrics	names starting `gen_ai_`

Connect the gateway in Settings > Integrations and CostGraph serves this target to the operator. Override the namespace or selector on the connection if you run it somewhere else.

How the metric maps

gen_ai_client_token_usage becomes costgraph_gen_ai_usage_total:

Envoy AI Gateway label	CostGraph label
`gen_ai_response_model`	`costgraph_model`
`gen_ai_token_type`	`costgraph_usage_kind`
`gen_ai_provider_name`	`costgraph_host`
`customer_id`	`costgraph_sub_account`

The unit is always tokens. Envoy AI Gateway also reports gen_ai_token_type="total". CostGraph reads the input and output series and drops the total, so the same tokens are never counted twice.

Attribute usage to your customers

Map a request header onto the metrics. This is a controller-wide setting, so every gateway that controller manages shares the mapping.

helm upgrade -i aieg oci://docker.io/envoyproxy/ai-gateway-helm \
  --version v1.1.0 -n envoy-ai-gateway-system --reuse-values \
  --set controller.metricsRequestHeaderAttributes="x-costgraph-customer-id:customer.id"

--reuse-values keeps the settings the release already has. Without it, an upgrade that passes only this one --set discards every other value you configured.

The attribute customer.id arrives as the label customer_id: dots become underscores on the Prometheus side.

Group usage by agent run

Add the run header to the same setting, so a run’s model cost appears next to its sandbox cost on AutoPilot > Sandboxes > Runs. Separate the mappings with a comma, escaped as \, for Helm:

helm upgrade -i aieg oci://docker.io/envoyproxy/ai-gateway-helm \
  --version v1.1.0 -n envoy-ai-gateway-system --reuse-values \
  --set controller.metricsRequestHeaderAttributes="x-costgraph-customer-id:customer.id\,x-costgraph-run-id:run.id"

Send the header x-costgraph-run-id on each request, set to the run tag of the run’s sandboxes. Refer to Group cost by run.

AI gateways

Connecting, pricing, and sending usage yourself.

​What CostGraph scrapes

​How the metric maps

​Attribute usage to your customers

​Group usage by agent run

AI gateways

What CostGraph scrapes

How the metric maps

Attribute usage to your customers

Group usage by agent run