Repository navigation
fix(openrouter): run Fable 5.2 at medium effort by default - #1147
Draft
sosidudku1 wants to merge 2 commits into
Draft
sosidudku1 wants to merge 2 commits into
sosidudku1 wants to merge 2 commits into
Conversation
The agent sends no reasoning effort unless a Fusion worker sets one, so Fable 5.2 on OpenRouter ran at the API default, high. On a measured task that cost about four times medium for the same tests passed. A turn that asks for no effort now sends medium for Fable 5.2 and later on OpenRouter; an explicit effort, extraBody or model params still win.
The risk reviewer and context compaction send disableThinking with a small output cap. Adding the Fable medium default there would turn thinking on inside that cap.
This branch has not been deployed
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
The agent only sends a reasoning effort when something asks for one (a Fusion worker setting, context compaction). A normal turn on Claude Fable 5.2 through OpenRouter therefore ran at the API's default,
high. Claude Code and the Claude apps run the same model atmedium.Now, when a turn asks for no effort and the provider is OpenRouter, Fable 5.2 and later get
reasoning: { effort: "medium" }. An explicit effort on the turn, anextraBody.reasoning(the documented workaround) or modelparamsstill win. Other models, older Fable versions and other provider kinds are unchanged.This is option 2 from the issue (a medium default). The effort switch in the settings (option 1, overlaps ATO-490 / ATO-606) is not part of this PR.
Why
On a measured task (59 hidden tests, 3 runs each) Fable 5.2 at
hightook 26.6 min and $7.70 per task; atmedium4.2 min and $1.87, with the same 59/59. Linear: ATO-1368.Open question: the direct Anthropic connection and the subscription CLI are not changed here. Whether Anthropic's OpenAI-compatible endpoint honours
reasoning_effortneeds a real key to confirm.How it was verified
model-params.test.ts:defaultReasoningEffortfor Fable 5.2+ and for models it must leave alone.openai-build-body.test.ts: medium on OpenRouter by default; turn effort,extraBodywin; other models and kinds untouched.Not yet checked on the second Mac.