Skip to content

feat(model-selector): add thinking-level quick selector in chat dropdown - #2050

Open
xielixing wants to merge 1 commit into
GCWing:mainfrom
xielixing:feat/model-thinking-level-1545
Open

feat(model-selector): add thinking-level quick selector in chat dropdown#2050
xielixing wants to merge 1 commit into
GCWing:mainfrom
xielixing:feat/model-thinking-level-1545

Conversation

@xielixing

Copy link
Copy Markdown

Summary

Adds a compact reasoning effort (thinking level) quick-selector at the bottom of the Flow Chat model selector dropdown.

When the currently selected model has thinking enabled, the dropdown shows a row of effort-level buttons that the user can click to instantly change the model's reasoning effort and persist it to config — without navigating to Settings.

Effort options per provider

  • Responses API (OpenAI): None, Minimal, Low, Medium, High, Extra High
  • Anthropic adaptive: Low, Medium, High, Max
  • DeepSeek: High, Max

Changes

  • ModelSelector.tsx: Add handleSetReasoningEffort handler (persists to �i.models config, emits mode:config:updated) and effort selector UI section in the dropdown

easoning.ts: Add getReasoningEffortOptionsForModel() helper that returns the correct effort options based on provider

  • ModelSelector.scss: Add styles for the effort selector section
  • i18n: Add
    one, minimal, max labels to
    easoningEffort block; add hinkingLevelUpdateFailed message in �n-US, zh-CN, zh-TW

Closes #1545

Add a compact reasoning effort selector at the bottom of the Flow Chat
model selector dropdown. When the currently selected model has thinking
enabled, the dropdown shows a row of effort-level buttons (None, Low,
Medium, High, Max, etc.) that the user can click to instantly change
the model's reasoning_effort and persist it to config.

Closes GCWing#1545
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.

[Feature]: 建议增加模型思考层级设置功能

1 participant