You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Adds a Vertex AI collector to the GCP provider, enabled via
`-gcp.services VERTEX`
Exports three metric types:
- cloudcost_gcp_vertex_token_input_usd_per_1k_tokens {model, region}
- cloudcost_gcp_vertex_token_output_usd_per_1k_tokens {model, region}
- cloudcost_gcp_vertex_instance_total_usd_per_hour {machine_type, use_case, region, price_tier}
Token metrics cover any model family billed through Vertex AI (Gemini,
Claude, etc.). SKUs with no regional breakdown fall back to
region="global". Compute metrics cover custom training and online
prediction nodes with on_demand and spot price tiers.
Pricing is fetched from the GCP Billing API at startup and refreshed
every 24 hours. Other Vertex AI products (embeddings, reranking,
AutoML, Vector Search) are not yet collected.
SKU description regex patterns must be verified against the live GCP
Billing API before production use.
Copy file name to clipboardExpand all lines: docs/metrics/gcp/vertex.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -33,4 +33,4 @@ Metrics exported for the GCP Vertex AI service.
33
33
34
34
## Notes
35
35
36
-
Pricing data is fetched from the GCP Billing API at startup and refreshed every 24 hours. SKU descriptions are matched using regular expressions; unknown SKUs are skipped. Verify SKU description patterns against the live Billing API when adding new Gemini models or machine types.
36
+
Pricing data is fetched from the GCP Billing API at startup and refreshed every 24 hours. SKU descriptions are matched using regular expressions; unknown SKUs are skipped. Verify SKU description patterns against the live Billing API when adding new models or machine types.
0 commit comments