CATCH #848
DETECTED AUG 19, 2026 22:12 UTC
THE CATCHscore 5/10
EXHIBIT 848
Aug 19, 2026 22:12 UTC
web app
https://platform.claude.com
Anthropic's Claude platform adds new observability features for output tokens, detailing internal reasoning token usage.
READY TO POST:
anthropic's claude console just got an update with more detailed output token breakdowns, looks like you can now see tokens spent on internal reasoning.
new strings in claude's platform hint at better observability for output tokens, specifically tracking how many tokens go towards a model's 'internal reasoning'.
EVIDENCE — the receipts
⚠️ FROM WEB-APP BUNDLE — may include vendored-dependency strings, verify before posting.
Anthropic Console (https://platform.claude.com) deployed new frontend code [8f9cc08c319110a9 -> a3cb7cd7215502ce]: +177 strings, -555 strings, 2 interesting
[feature_flag] (2)
+ Breakdown of output tokens by category.
`output_tokens` remains the inclusive, authoritative total used for billing.
This object provides a read-only decomposition for observability — for example,
how many of the billed output tokens were spent on internal reasoning that may
have been summarized be
+ Number of output tokens the model generated as internal reasoning, including
the thinking-block delimiter tokens.
Reflects the raw reasoning the model produced, not the (possibly shorter)
summarized thinking text returned in the response body. Computed by
re-tokenizing the raw reasoning text, so it
caught by the 20-minute sweep ·
the live wire · all receipts