Where the AI budget went: spend, model calls and tokens by feature, agent, model and day, at /claw/usage. Admins only.
The Usage page
Talon Claw → Usage (/claw/usage) is for admins; other roles are shown an access-denied page. Choose Today, 7 days, 30 days (the default) or 90 days at the top.
Spend
Estimated spend over the period. Calls no price covers (self-hosted models and the subscription engines) are counted but not priced, so the total then reads as a floor (≥ $…), or unpriced when no call has a price.
Model calls
How many calls Studio made to a model.
Tokens in / out
Input and output tokens.
Cache reads
The share of input tokens served from the prompt cache.
Spend by day
A bar chart of priced spend per day.
By feature, By agent, By model
Spend and calls ranked by what made the call, which agent it ran for, and which model answered.
Budget
Tokens and cost used against the org's budgets when one is set, with a reminder that the window rolls; otherwise a note that no budget is set.
Talon Claw → Usage over 7 days: the spend, call, token and cache tiles, spend by day, and spend ranked by feature, agent and model, with the budget beneath.
From 80% of the budget
Once 80% of the token or cost budget for the window is spent, the page says so in an amber banner, and every agent is told to be economical: fewer and smaller tool calls, no speculative exploration. At 100% new AI requests are refused until usage ages out of the window, or until an admin raises the budget under Talon Claw → Settings.
What is counted
Every model call Studio makes is recorded under a feature: chat, chat summaries, agent runs, automation steps, automation authoring (Build with AI), stream classification, SIEM migration, Sigma conversion review, ATT&CK mapping, threat-intel extraction, synthetic events, memory upkeep, search intent, commit messages, connection tests, run reviews and agent improvement proposals. Each one counts toward the token and cost budgets. Usage recorded before v1.0.2-beta reads Other (unlabelled).
The 30-day summary
The Usage tab of Talon Claw → Settings keeps a shorter view: requests, tokens, cache hit rate and estimated cost for the last 30 days, by model.
