🤖 fix: update Claude Opus 4.5 thinking policy to match API #752
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Claude Opus 4.5 uses the effort parameter which only supports low/medium/high (no 'off' option). Update thinking policy to reflect this: - Return ['low', 'medium', 'high'] for opus-4-5 models - Fallback to 'high' (instead of 'medium') when 'off' is requested - Add comprehensive test coverage for Opus 4.5 policy - Add Storybook story showcasing 3-position slider This aligns the UI with the Anthropic API's capabilities, similar to how Gemini 3 is handled with ['low', 'high']. Fixes the issue where users could select 'off' for Opus 4.5, which would result in no effort parameter being sent to the API. _Generated with `mux`_ Change-Id: If402fe10a6061ce21dac4eb23a29ca58a9ca3613 Signed-off-by: Thomas Kosiewski <tk@coder.com>
Change-Id: Idfbb6de0240ee4a3f82835bb3cc68469b1eeb7c0 Signed-off-by: Thomas Kosiewski <tk@coder.com>
The play function was causing test failures. The story is primarily for visual documentation of the 3-position slider, so the play function isn't necessary. Change-Id: I1b0123cecf1b1bba65abd759fd68bd63333ee2d3 Signed-off-by: Thomas Kosiewski <tk@coder.com>
The InteractiveDemo play test was failing because it shared persisted state with other stories (via the shared 'storybook-demo' workspaceId). When DifferentModels runs first and sets thinking to 'medium' for Opus 4.5, the InteractiveDemo story inherits that state. Fix by giving InteractiveDemo its own unique workspaceId to ensure test isolation. Change-Id: I31bc0c6a2fc21a9a0084b966ce047422ce91339d Signed-off-by: Thomas Kosiewski <tk@coder.com>
ThomasK33
added a commit
that referenced
this pull request
Nov 25, 2025
Claude Opus 4.5 uses the effort parameter which only supports `low`, `medium`, and `high` (no 'off' option). This PR updates the thinking policy to reflect the API's capabilities. ## Changes - **Policy Update**: Return `['low', 'medium', 'high']` for `opus-4-5` models (no 'off') - **Fallback Logic**: When 'off' is requested, fallback to 'high' for best reasoning experience - **Test Coverage**: Added comprehensive tests for Opus 4.5 policy (15 tests pass) - **Storybook**: Added `Opus45ThreeLevels` story showcasing the 3-position slider ## Impact | Before | After | |--------|-------| | 4-position slider (off/low/medium/high) | 3-position slider (low/medium/high) | | Could select 'off' (invalid for API) | Cannot select 'off' (matches API) | | Invalid UI state possible | UI always matches API capabilities | ## Testing - ✅ All 419 browser utils tests pass - ✅ All 15 policy tests pass - ✅ TypeScript typecheck passes - ✅ Opus 4.5 fallback to 'high' works correctly This aligns Opus 4.5 with how Gemini 3 is handled (which has 2 levels: low/high). _Generated with `mux`_ --------- Signed-off-by: Thomas Kosiewski <tk@coder.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Claude Opus 4.5 uses the effort parameter which only supports
low,medium, andhigh(no 'off' option). This PR updates the thinking policy to reflect the API's capabilities.Changes
['low', 'medium', 'high']foropus-4-5models (no 'off')Opus45ThreeLevelsstory showcasing the 3-position sliderImpact
Testing
This aligns Opus 4.5 with how Gemini 3 is handled (which has 2 levels: low/high).
Generated with
mux