PIPEROLL / SEISMOGRAPH
Published finding

deepseek-reasoner@deepseek: thinking-token use rose on the verbosity probe set

Movement on the daily battery. This model measured against its own witnessed baseline.

90 tok→951 tok
modeldeepseek-reasoner@deepseek
identityalias · a floating name the provider may re-point; a change here means what is served under this name changed
dimensionVerbosity probes
metricthinking_tokens_mean
decision ruleRelative shift past the serving-tripwire threshold
first flagged2026-09-10 · still flagged 2026-09-15 (6 readings)
coverage1 call error on this reading

Pattern: thinking-token use moved while the pass rate on the same probes stayed flat (1.00 before, 1.00 at onset). That is the signature of a serving-side change - an effort default or a routing change - rather than a different model behind the name, which usually moves accuracy too. Stated as a pattern, not a cause.

Dated records alongside this reading

An annotation attaches a dated external record (a provider changelog entry, a documented deprecation, a known configuration artefact) to a reading the instrument already produced. It changes no reading and establishes no cause; the instrument detects, attribution stays a human judgment.

Provider record on the onset date · 2026-09-10

DeepSeek's API updates page dates the DeepSeek-V4.1-Flash release and its API price change to 2026-09-10, and states that V4 Pro continued to be served unchanged after 2026-09-14. The onset coincides with the V4.1 Flash release day. It does not coincide with a Pro routing change (none occurred on that date, per the same page), so an earlier working hypothesis that this was a Pro-to-Flash reroute was wrong. What changed behind the deepseek-reasoner alias on 2026-09-10 is not established by the instrument.

Annotation added 2026-10-21.

Trace the evidence

The witnessed reading for this date and battery, and the Rekor witness bundle. The full series is in series.json.

This entry records an observation under the stated decision rule. Watches remain unconfirmed; latency is operational context. The instrument does not establish the cause of a change.