Anthropic Claude API cost tracking and savings.
What Audon reads. And what it never reads.
Reads
- Daily cost report by workspace and description (model, token type)
- Daily message token usage by model, workspace and API key
- Claude Code usage per person and model (the Claude Code share of the same spend)
Never reads
- Prompts or responses
- Priority Tier costs (not in the cost report)
- Claude.ai seat analytics (Claude Enterprise connector)
The key to create. Used read-only, encrypted, never shown again.
- Credential
- Admin API key (sk-ant-admin…)
- Where to create it
- platform.claude.com → Settings → Admin keys. Requires the organization admin role. Claude Enterprise (claude.ai) organizations use the Claude Enterprise connector.
- Provider documentation
- platform.claude.com/docs/en/manage-claude/usage-cost-api
- Step by step
- Provider setup in the docs
Savings levers on Anthropic (Claude API) data. Sized on your own spend, each with the exact fix.
Pay the fast or priority surcharge only where someone waits; run work nobody waits on at flex or batch rates.
Drop the regional or US-only processing surcharge where no contract requires it.
Move to a newer, cheaper version of the same model, or route simpler requests to a cheaper model.
Run tests, evals and CI on a cheaper model, a smaller sample or a nightly schedule.
Stop resending growing history and stay under long-context price thresholds.
Cap outputs and set reasoning effort per task, so you pay for the answer, not the padding.
Serve repeated prompt prefixes from cache, and stop paying for cache writes that are never read.
Give each person the cheapest plan that covers how they work: remove, downgrade, upgrade or move to a seat.
Default coding agents to a cheaper model or the tool’s own model pool; keep the premium model for hard tasks.
A lever writes a potential saving. A change you mark as done is measured from billing; an approved experiment makes it a verified saving.
Good to know.
- Standard and workspace keys are rejected: the key must be an Admin key (sk-ant-admin…) from an organization admin.
- Priority Tier spend isn’t in Anthropic’s cost report. The default workspace appears without an id.
- Claude Enterprise organizations on claude.ai use the Claude Enterprise connector instead.
Questions about Anthropic (Claude API).
What does Audon need to connect Anthropic (Claude API)?
What does Audon read from Anthropic (Claude API)?
Does Audon read prompts, responses or code?
Which savings levers use Anthropic (Claude API) data?
How do I revoke access?
Last updated
Other integrations.
- What we read
OpenAI
Costs and usage by project, model and line item, read with an organization Admin key.
- What we read
Claude EnterpriseNew
Per-person activity, idle seats, and daily cost and tokens by product and model.
- What we read
Claude CodeNew
Tokens and cost per person and model, from the Admin report or Claude Code’s own telemetry.
- What we read
CodexNew
Tokens per day and model from Codex’s own OpenTelemetry export, with no key stored.
- What we read
xAINew
Daily team spend by model and endpoint, read with a team-scoped management key.
- What we read
CursorNew
Members, usage per person and model and on-demand spend, or the usage export on individual plans.
- What we read
DevinNew
Daily ACUs by product and per person, priced with your own price per ACU.