Follow Discord
Sweep 22 Oct 2026 · 04:32Z Build v2.1.308 527 read Stable v2.1.308 Latest v2.1.308 Next v2.1.308 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.300 ·

Offer to choose an auto-compact size for 1M-context sessions

Users with a one-million-token context can be offered a choice of when auto-compact kicks in, from 300k to 900k tokens

You'll notice Improvements
JSON All of v2.1.300
You'll noticeTier: how much it should matter to you
3Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
CompactionArea: what it touches
ImprovementsKind: in v2.1.300,
ImprovementsSection of the release
What

The context window is how much of the conversation the model can hold at once, measured in tokens (small chunks of text). Auto-compact summarises older parts of the conversation to free up space. Users with a one-million-token context window can now be offered a choice of window size for auto-compact, anywhere from 300k to 900k tokens.

The offer is not made when the size is already set some other way:

  • by an environment variable
  • by an experiment or a server setting
  • because auto-compact is turned off

It is controlled by tengu_binary_music, a setting sent from Anthropic's servers. When the server sends nothing, the offer stays off.

Why

This is a trial of letting people pick when compaction happens in long-context sessions, rather than leaving it fixed.

See this entry in the whole of v2.1.300 →

Feedback