{"version":"2.1.319","anchor":"autocompact-put-off-when-the-model-window-is-much-larger-tha","canonical_anchor":"autocompact-put-off-when-the-model-window-is-much-larger-tha","heading":"Automatic compaction can wait for the model's full context window","tier":"notice","area":"Compaction","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319\/e\/autocompact-put-off-when-the-model-window-is-much-larger-tha","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319","markdown":"### Automatic compaction can wait for the model's full context window\n\nAutocompact can now hold off at its usual point and compact at the model's whole-window threshold instead\n\n**Unclear.** Whether this later compaction is active by default is not stated.\n\n**What**\n\nAutocompact is when Claude Code automatically summarizes a long conversation to free up room. It can now decline to compact at its usual threshold. This happens when no prompt since the last summary was under a configured size; the conversation then compacts at the threshold of the model's whole context window (the total amount of text the model can hold) instead.\n\n**Why**\n\nOn models with a large context window, compaction may happen later than before.\n\n- Area: Compaction\n- Tier: You'll notice\n- Useful: 3\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}