Skip to content

Apply the phone's reasoning effort for Autohand AI models - #769

Open
igorcosta wants to merge 1 commit into
mainfrom
fix/mobile-set-model-effort
Open

igorcosta wants to merge 1 commit into
mainfrom
fix/mobile-set-model-effort

Conversation

@igorcosta

Copy link
Copy Markdown
Contributor

The Autohand Code iPhone app now lists the CLI's Autohand AI catalogue (Moa, Fantail, Auto and the rest, from src/providers/models.json) with each model's effort levels. Its set_model action sends reasoningEffort, which the relay dropped.

  • MobileRelay passes reasoningEffort to the model change handler when present.
  • applyModelChangeRemote stores it for autohandai the way /model does: Moa keeps medium, high or xhigh; a model without effort levels (Fantail) drops a stored effort; an effort the model doesn't offer is ignored; an effort-only change applies without a model change.

Tests: tests/mobile and tests/core/agent pass (1493).

… model

set_model from the paired phone now carries reasoningEffort. For provider autohandai it is
stored the way the /model picker stores it: Moa keeps medium, high or xhigh, a model without
effort levels drops it, and an effort-only change applies without a model change.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant