{"version":"2.1.284","anchor":"persisted-cost-state-now-rejects-implausible-per-model-token","canonical_anchor":"persisted-cost-state-now-rejects-implausible-per-model-token","heading":"Saved cost data with impossible token totals is now rejected","tier":"internal","area":"Elsewhere","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284\/e\/persisted-cost-state-now-rejects-implausible-per-model-token","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284","markdown":"### Saved cost data with impossible token totals is now rejected\n\nSaved cost data is now rejected when its token totals across models are larger than the limit allowed for a single number\n\n**Unclear.** It is not stated what happens to the saved cost data when it fails this check, or what the limit is.\n\n**What**\n\nClaude Code saves a record of cost and token usage for each model. When that record is read back, it is now checked more strictly. For input tokens, output tokens, cache-read tokens and cache-creation tokens, the total across all models must not be larger than the limit already applied to each single number. A saved record that breaks this is treated as invalid.\n\n**Why**\n\nThis protects the figures shown by `\/cost` from being filled with corrupted or overflowing totals from a damaged saved record.\n\n- Area: Elsewhere\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 1\/5"}