Unclear Whether this setting is currently visible or exposed for a given account isn't settled here, since a related unreviewed flag exists (tengu_jazzy_goblet, currently reading off for this site's account) that isn't confirmed to control this specific setting.
What
- The
/configsettingswitchModelsOnFlag(whether Claude Code automatically switches to a fallback model when the current model refuses/flags a request) is now an explicit enum of 'switch automatically' or 'ask each time', rather than a plain boolean, though it still accepts true/false/on/off/yes/no from the CLI or config files. - When a model refuses or flags a request and Claude Code wants to switch to a fallback model, it can now ask interactively with a question like "Switch to ... and keep going whenever this happens?", offering to switch automatically or pause, and remembers the choice.
- Changing this setting now fires a
tengu_refusal_fallback_setting_changedtelemetry event and persists the change to user settings. - New telemetry events were registered:
tengu_refusal_fallback_preference_prompt, plustengu_policy_limits_cache_state_at_first_promptandtengu_policy_limits_fetch.
Why This gives users more explicit control over automatic model fallback on refusal, letting them decide in the moment (and remember the choice) rather than relying on a single fixed setting.
tengu_jazzy_goblet Off in both readingsThe flag server returned off for the account this site reads and for the anonymous baseline. A reading of off cannot rule out a rollout these two readings sit outside of.
This account: off · anonymous baseline: off · compiled default in v2.1.269: not a boolean we can read
Read once, for one account on one subscription tier, against v2.1.269. It isn't a statement about your account. What a flag value here can and cannot tell you
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
| [`switchModelsOnFlag`](#switchmodelsonflag) | Switch models automatically or pause when a [safety classifier](/docs/en/model-config#ask-before-switching) flags a request | Model and responses | Any file |settings-reference see the edit
Whether this setting is currently visible or exposed for a given account isn't settled here, since a related unreviewed flag exists…
Anthropic's documentation has since written up switchModelsOnFlag, on All settings.
New in this build: tengu_jazzy_goblet