Follow Discord
Sweep 27 Oct 2026 · 19:04Z Build v2.1.312 531 read Stable v2.1.308 Latest v2.1.312 Next v2.1.312 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.312 ·

How often Claude Code double-checks reused request data can now be tuned remotely

The sampled check that reused request data is correct now runs at a rate Anthropic can scale from its servers

Under the hood Internal Changes
JSON All of v2.1.312
Under the hoodTier: how much it should matter to you
1Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
API RequestsArea: what it touches
Internal ChangesKind: in v2.1.312,
Internal ChangesSection of the release
What

When Claude Code builds a request to send to the model, it can reuse parts of a request it has already put together. Now and then it checks that the reused data matches what it should be, and catches wrong data before sending. How often that check runs is now scaled by a number set remotely by Anthropic, between 0.1 and 1. Any value outside that range counts as 1. Whether the first check happens in a session is now decided once per session, rather than by fixed limits.

Why

Anthropic can now raise or lower how often these checks run without shipping a new version. Nothing you do changes.

See this entry in the whole of v2.1.312 →

Feedback