Anthropic APINew featureinfo

Anthropic Models API now reports whether a model allows disabling thinking

Detected
Oct 6, 2026
Confidence
85%

What changed

Anthropic added documentation for a new thinking.types.disabled field in the Models API capabilities object, which indicates whether a model accepts thinking: {type: "disabled"} to turn off thinking mode. This includes clarification that supported: false means the model rejects the disabled setting with a 400 error.

What to do

Developers working with thinking-capable models can now programmatically check capabilities.thinking.types.disabled.supported via the Models API to determine if disabling thinking is allowed, rather than inferring it from the model ID.

Affected

model: capabilities.thinking.types.disabledendpoint: GET /v1/models

Source: docs.anthropic.com (Anthropic's own page, checked by Breakwatch)

Is your code affected?

Search your repositories for the Anthropic SDK and the names above:

npm: @anthropic-ai/sdkpip: anthropicruby: anthropicgo: github.com/anthropics/anthropic-sdk-gojava: com.anthropic:anthropic-javarust: anthropic

Or let Breakwatch do it: connect a repo and see which files call Anthropic, then get the next change as an alert pointing at those files. Check a repo free

Get Anthropic API changes by email

Once a week: the breaking changes and upcoming deadlines for Anthropic, each linked to the vendor's own page. Free, no account. One-click unsubscribe.

More Anthropic API changes

Every Anthropic API change →