All issues

quality strategy tiebreaker favors expensive models, contradicting balanced strategy's "cheapest wins"

#116

· xiaomingwo opened

on Sep 14, 2026

Streaming requests don't cascade mid-flight when primary fails after first chunk

#115

· hunterwhip opened

on Sep 14, 2026

Feature: Add per-provider weight/priority configuration for routing decisions

#114

· leomete opened

on Sep 14, 2026

Sibling version deduplication breaks when canonical_model_base strips too aggressively

#113

· danielbennett888 opened

on Sep 14, 2026

Question: How does model="auto" interact with LiteLLM's internal retry loop?

#112

· miatco opened

on Sep 14, 2026

balanced strategy drops ALL models when quality_scores dict is empty due to AA fetch failure

#111

· jamesleanr opened

on Sep 14, 2026

Router timeout=30s is hardcoded and ignores the router_cooldown_seconds setting

#110

· hrobers3 opened

on Sep 14, 2026

model="auto" with preferred_models silently ignores non-deployable entries instead of erroring

#109

· najwac opened

on Sep 14, 2026

fix(cache): prompt cache serves wrong responses — key omits sampling params, omitted temperature treated as deterministic

#84

· hasitpbhatt opened

on Aug 21, 2026

fix(analytics): every streaming request logs latency_ms=0, corrupting latency percentiles

#82

· hasitpbhatt opened

on Aug 21, 2026

fix(observability): unhandled exceptions are swallowed with zero logging (500s and auth 503s undebuggable)

#80

· hasitpbhatt opened

on Aug 21, 2026

fix(auth): provider keys sealed with a publicly-known dev key by default — fail closed at startup

#76

· hasitpbhatt opened

on Aug 21, 2026