Monitoring changedcowork/monitoring
Nearest release: v2.1.310, published an hour before upstream edited the page. Shown because the two are within 24 hours of each other. Nothing here says the release caused the edit.
Upstream edited this page at 23 Oct 2026 22:43 UTC, give or take a minute or two: the time comes from Anthropic’s own sitemap rather than from a commit. This site recorded the change at 23 Oct 2026 23:07 UTC.
Upstream edited
Recorded here
Lines+4added
Lines−0removed
From line
169
where the diff opens
First seen
14 Aug 2026
this site's first read of the page
Recorded edits11to this page, all time
The whole hunk
from line 169, old and new numbered
/
from line 169
169169| `output_tokens` | Number of output tokens |
170170| `cache_read_tokens` | Number of tokens read from cache |
171171| `cache_creation_tokens` | Number of tokens used for cache creation |
172| `cache_creation_5m_tokens` | Portion of `cache_creation_tokens` written with a 5-minute cache lifetime |
173| `cache_creation_1h_tokens` | Portion of `cache_creation_tokens` written with a 1-hour cache lifetime |
172174| `speed` | `"fast"` or `"normal"` |
175
176Claude Desktop 3.1623.0 and later send `cache_creation_5m_tokens` and `cache_creation_1h_tokens`. Both are `0` when the API response doesn't report the split, even when `cache_creation_tokens` is above zero. If your deployment uses an [LLM gateway](https://claude.com/docs/third-party/claude-desktop/gateway.md), confirm that the gateway returns the `cache_creation` object in each response's `usage`.
173177
174178### API error event
175179
No line in this hunk matches that.