AI usage

When you would want it

See where your own provider spending goes and how much of your included Tagflow allowance remains.

Where to find it

• All Features → AI usage
• Show usage

What you get out of it

• See where the money goes, model by model, before your provider's bill arrives.
• Spot the agent run or long chat that used far more than you expected.
• See your real pattern of AI use without collecting new activity data or sending local history to a server.
• Honest numbers: estimates are marked, and a model with no known price says No price instead of showing $0.
• Your own provider history stays local. Included Tagflow usage counts are stored on the server to enforce the shared account allowance.

How to use it, step by step

1. Select Usage in the chat's header, or Usage by model in the context meter's menu, or run AI: Show Usage.
2. Pick a range: Today, 7 days, 30 days or All time.
3. Read the totals - tokens in and out, requests, and the estimated cost - and the bar per day underneath.
4. Activity shows the past year independently of the totals range: your current streak, longest streak and active days. A day is active when the chat or an agent makes an AI request on this computer.
5. Hover, click or focus a calendar square for its date, requests and tokens. Arrow keys inspect neighbouring days; Home and End go to the oldest and newest day. Small windows scroll the calendar horizontally, starting at the recent days.
6. Streaks follow local calendar days. Yesterday keeps a streak alive while today is still empty; the longest streak is measured within the displayed year.
7. Look down the list of models: each shows its share of tokens, its requests from the chat and from agents, and its cost at the catalogue's price.
8. A ≈ marks counts Agentfly estimated because the provider did not report them. Clear usage removes this machine's history, activity and streaks.
9. For included Tagflow AI, read your account-wide allowance above the history. It includes active reservations, resets at midnight UTC, and cannot be reset by clearing local history.

Why this beats the usual way

Provider dashboards show one total a day late and cannot tell your chat from your agents. Agentfly counts every request as it happens, split by model and by where it came from, right inside the editor.

Find it in Universal Search

AI usage · The assistant · usage · tokens · token count · cost · spend · spending · billing · how much · price · per model · budget · activity · streak · heatmap · calendar · active days