{"version":"2.1.296","anchor":"token-count-request-builds-thinking-config-from-a-helper","canonical_anchor":"token-count-request-builds-thinking-config-from-a-helper","heading":"Token counting now matches each model's thinking settings","tier":"notice","area":"Elsewhere","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.296\/e\/token-count-request-builds-thinking-config-from-a-helper","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.296","markdown":"### Token counting now matches each model's thinking settings\n\nRequests that count tokens now set the thinking option to suit the model instead of always using one fixed setting\n\n**What**\n\nClaude Code sometimes asks the API how many tokens a conversation uses. Tokens are the units the model reads text in. These requests include a setting for thinking, the model's step-by-step reasoning before it answers. That setting used to be fixed: thinking switched on, with a set token budget. It is now chosen for the model in use, so models that handle thinking differently get a setting that fits them.\n\n**Why**\n\nToken counts should be more accurate on newer models, and count requests are less likely to send a thinking setting that a model does not accept.\n\n- Area: Elsewhere\n- Tier: You'll notice\n- Useful: 1\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}