Repository navigation
fix(combo): model-scoped 5xx on passthrough providers (upstream candidate) - #2
Open
kenny000666 wants to merge 6 commits into
Open
kenny000666 wants to merge 6 commits into
kenny000666 wants to merge 6 commits into
Conversation
…ibling models (#kenny-fix)
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Upstream PR cheaper-inference#3637 introduced exhaustedConnections tracking to skip connections returning 502/503/504. This operates at connection level, not model level. For passthrough providers (e.g. ProfundoAI) exposing multiple models via one baseUrl+apiKey, a 503 on one model (kimi-k3) poisons the whole connection — sibling models (claude-opus-5, gpt-5.6-sol) get skipped in combo fallback.
Fix
Added hasPerModelQuota() guard in open-sse/services/combo/targetExhaustion.ts that suppresses the connection-level skip for 502/503/504 when provider has per-model quota. Error is still recorded but sibling models remain eligible for fallback.
Commits
Verification
Deployed to prod (ghcr.io/kenny000666-apps/omniroute:3.8.52) 2026-10-27. Pod omniroute-84c66f5759-m62pv healthy, combo routing succeeding with 0 fallbacks.
Upstream Path
After 1-2 weeks stable in prod, propose to cheaper-inference/OmniRoute against next release branch.