Sign in required
This dashboard uses your Siemens account.
Sign in with MicrosoftDaily spend against baseline
actual baseline (all premium)
By model
By tier
By tool
Create a key
A key reaches this gateway and nothing else. It replaces the AWS credentials that would otherwise sit in your Claude Code settings file, where every subprocess inherits them.
Your keys
Team usage
Team level is the default view. Per-person spend is performance-adjacent data, so it needs a reason, not curiosity.
By model
By tier
Candidate health
Model catalog
Quality scores come from Artificial Analysis. They are an external, provider-blind prior: they describe the model, not our Bedrock path. Ranking is coding index per 1000 USD of a fixed workload, because per-token price ignores how many tokens a model actually burns.
Routing table
Tiers describe what a request demands, not what you are doing. A candidate is model plus reasoning effort plus wire protocol plus region, because all four change the bill.