Fetching from the wire…
Tools2026-08-27 · source-backed
The same 2.1.243 release lets API-key and cloud-provider users hold a 1-hour prompt cache on the main conversation while subagents stay at the 5-minute default (Releasebot). That split is right for orchestration patterns where a long-lived parent context is expensive to rebuild but short-lived fan-out subagents would only pay the 1-hour cache-write premium without ever reusing it. A modelPricing managed setting also lets an organization's contracted rates feed /cost, the status line and telemetry instead of list price.
Each link below shares sources, entities, or timing with this story.
Shared entity: Releasebot / Same source / Shared topic / Earlier coverage / Tension
Both cover Releasebot; cite the same source (Releasebot); overlapping topics (context, cost).
Shared entity: TTL / Shared topic / Earlier coverage
Both cover TTL; overlapping topics (between, cache, contracted, conversation, cost); earlier TTL coverage from 2026-08-25.
Shared entity: Releasebot / Same source / Earlier coverage / Tension
Both cover Releasebot; cite the same source (Releasebot); earlier Releasebot coverage from 2026-08-04.
Both cover Releasebot; cite the same source (Releasebot); earlier Releasebot coverage from 2026-07-13.
Both cover Releasebot; cite the same source (Releasebot); earlier Releasebot coverage from 2026-06-18.
Shared entity: TTL / Shared topic / Earlier coverage
Both cover TTL; overlapping topics (between, cache, context, cost, hour); earlier TTL coverage from 2026-08-15.
Shared entity: Releasebot / Same source / Earlier coverage
Both cover Releasebot; cite the same source (Releasebot); earlier Releasebot coverage from 2026-08-16.
Both cover Releasebot; cite the same source (Releasebot); earlier Releasebot coverage from 2026-08-09.