See what a conversation used

Before you start

• Use the Enjoy app on your computer. Usage is kept on the computer where the agent ran, so the web and iPhone apps don’t show it.
• Use Claude Code. Other agents don’t report what each turn used yet.
• Send a message. Enjoy starts counting a conversation’s turns from the first message you send after updating, so older conversations show usage once you send another message.

Steps

1. Open the conversation’s ⋯ menu

   At the right end of its title, select the ⋯ button (Conversation actions). Under the title, Usage shows how many tokens the conversation has used, such as 3.6M tokens.

   1

2. Select Usage

   Every number the agent reported appears in four groups: This conversation, Last message, Right now (the conversation’s size and how much the model can hold), and Your plan (your 5-hour and weekly limits). In the first two, Tokens gives the total, with what it’s made of under it.

   2

3. Point to a number to see what it means

   The box under the numbers explains the one you point to or select. Until then, it reminds you that most tokens are cheap re-reads of the conversation.

   3

4. To keep the context in view, select Show in conversations

   It’s under Right now. From then on, every conversation shows how full its context window is beside its ⋯ button, such as 12%.

   4

5. Select the percentage to see the numbers

   It opens the conversation’s size and its context window. To hide the percentage again, clear Show in conversations there or in the ⋯ menu.

   5

Good to know

• Tokens are small pieces of text, about three quarters of a word each. Every step the agent takes sends it the whole conversation again, so most of a conversation’s tokens are Re-read from cache. Re-reading counts about a tenth as much as new text, which is why a large number there is usually fine.
• Conversation size is how much the agent re-reads at every step right now. The longer it gets, the more each message uses. A new conversation starts small again. See Use one conversation per task.
• API price is what the tokens would cost at the provider’s API prices, as the agent reports it. Your plan doesn’t charge per message; it’s there for comparison.
• Model requests and Time working count from when Enjoy started recording the conversation’s turns. Point to them to see the date.
• Last message shows its tokens only when Enjoy knows what the conversation had used before it. For a conversation from before you updated, they appear from the next message on.
• Your plan covers your whole account, every conversation and device included. See Check how much usage you have left.
• Show in conversations is off until you turn it on. It’s one setting for every conversation on this computer, and it shows only for conversations whose agent reports its context window.
• Deleting a conversation also deletes its usage.