Follow Discord
Sweep 25 Sep 2026 · 19:33Z Build v2.1.283 504 read Stable v2.1.274 Latest v2.1.283 Next v2.1.283 Feeds RSS JSON llms.txt llms-full.txt Unofficial
HomeReleasesv2.1.281entry #68

Claude Code v2.1.281 ·

Built-in model entry no longer forces thinking on and needs 2.1.280

One model in the bundled model selector config drops always_on thinking and now requires Claude Code 2.1.280

Group of 2 You'll notice Entry
JSON All of v2.1.281
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
EntryKind: in v2.1.281,
ImprovementsSection of the release

What

Claude Code ships with a built-in list of models, the model_selector_config, that feeds the model picker. It includes entries such as claude-fable-5-1. One entry in it changed. That entry has a hard_limit of 950000 and supports fast mode.

  • Its thinking option loses always_on: !0.
  • It gains min_claude_code_version: "2.1.280". Other entries carried "2.1.251".
  • It gains voice_model: "claude-opus-5".

Thinking is the step-by-step reasoning a model can do before it answers.

Why

In practice, thinking is no longer forced on for that model's picker entry, and the entry only applies to Claude Code 2.1.280 or later.

How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtThe finding does not say what dropping `always_on` or adding `voice_model` changes in what a reader sees or can choose.

See this entry in the whole of v2.1.281 →

Feedback