Glossary · term27 of 202
cache read
A token served from a prior prompt's cached context instead of being processed fresh, billed at a fraction of the base input price. Most tokens in a long agent session are cache reads, not new context.
also written as cache hit · cache reads · cached input
Related terms
JSONLA text file format storing one JSON object per line rather than a single nested document, making it easy to append events without rewriting the whole file.view termTTLThe time a cached prompt stays valid before it expires and must be rewritten at full cost. Shortening it turns yesterday's cheap cache reads into today's expensive cache writes.view termbilled iterationOne individually charged request within a multi-step agent run, including hidden steps like compaction, that top-level usage summaries can omit entirely.view termcache hit ratioThe share of total tokens in a session that came from cache reads rather than fresh input, which matters because cached tokens bill far cheaper than uncached ones.view termcache missA request whose prompt prefix no longer matches what was cached, forcing the full prefix to be billed again at the cache-write rate instead of the cheaper cache-read rate.view termcache writeThe cost of storing a prompt's context so later requests can read it from cache instead of reprocessing it. Billed above base input price, with the rate depending on how long the write stays valid.view termccusageA community tool that estimates API-equivalent dollar costs from Claude Code transcript files by converting logged token counts using published pricing.view termcontext-tier boundaryA pricing threshold where a task's context size crosses into a different billing tier, which can make measured per-request token savings disappear once a task grows past it.view termmessage_stop eventThe event in Claude Code's transcript format that would carry the final output token count for a response, but is never written, causing recorded output tokens to undercount actual billable output.view termprefix stabilityKeeping the leading portion of a prompt byte-identical across turns so it keeps qualifying for cheap cache reads instead of triggering an expensive cache rewrite.view termprompt cachingA billing mechanism where a stable, byte-identical prompt prefix gets reused across turns at a fraction of normal input cost, while any edit to that prefix forces a full-price rewrite.view term
At a glance
- cited by
- 3 posts
- categories
- 2
- first used
- jul 2026
Claude Code vs Codex Limits, Measured in the Same UnitClaude Code vs Codex limits, measured across 28 days and 56.8 billion tokens. Why token counts cannot predict quota burn, and what to read on each side instead.
Token-Compression Skills Do Not Survive MeasurementToken compression skills report savings the invoice never shows. Cache reads carry 87 percent of an agent bill, and cutting tokens can push the cost up.
Ways to Make the Most Out of Claude Fable 5 (2026)Get more from Claude Fable 5: pick the right effort level, plan with Fable while cheaper models implement, and claw back costs with caching and batching.