{"version":"2.1.295","anchor":"new-error-text-when-the-output-limit-is-hit-during-reasoning","canonical_anchor":"new-error-text-when-the-output-limit-is-hit-during-reasoning","heading":"New recovery message when output runs out while Claude is still thinking","tier":"notice","area":"Output Limits","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.295\/e\/new-error-text-when-the-output-limit-is-hit-during-reasoning","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.295","markdown":"### New recovery message when output runs out while Claude is still thinking\n\nIf Claude hits its output limit while still reasoning, it is now told to reason much more briefly and continue\n\n**What**\n\nClaude Code has a new message for when Claude reaches its output limit, the most text it may produce in one response, while still reasoning, before it has written any answer or used any tool. The message, beginning \"Output token limit hit while you were still reasoning\", tells Claude to reason much more briefly and carry on.\n\n**Why**\n\nIt changes how Claude Code recovers when a response is cut off during the thinking stage, steering Claude toward finishing the task rather than stalling.\n\n- Area: Output Limits\n- Tier: You'll notice\n- Useful: 3\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}