{"version":"2.1.301","anchor":"token-spend-tracked-separately-from-model-usage","canonical_anchor":"token-spend-tracked-separately-from-model-usage","heading":"Claude Code keeps a running count of tokens spent in a session","tier":"internal","area":"Usage & Limits","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301\/e\/token-spend-tracked-separately-from-model-usage","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301","markdown":"### Claude Code keeps a running count of tokens spent in a session\n\nClaude Code now keeps its own running total of input, output and cache tokens spent in a session, separate from model usage\n\n**What**\n\nTokens are the small chunks of text that models read and write, and they are what usage is measured in. Claude Code now keeps a separate running total of the tokens a session has spent, alongside its existing record of model usage. The total counts four kinds:\n\n- Input tokens\n\n- Output tokens\n\n- Tokens read from the cache\n\n- Tokens written to the cache\n\n**Why**\n\nThis is a building block. A separate token count could later support limits or reports based on tokens. Nothing you see changes because of it on its own.\n\n- Area: Usage & Limits\n- Tier: Under the hood\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}