#30484 LiteLLM Stability Sprint Roadmap
Hi all, we are treating stability as a first class citizen. Sharing our roadmap for bug fixes and stability improvements planned over this week. Please comment what bugs you'd like to see fixed over this sprint
## AI Engineering - LLM Translation
- [x] `/v1/model/info` returns inconsistent responses for team-based models, so team-scoped model metadata needs to be reconciled. - [ ] Container/code-interpreter file retrieval on Azure fails for service-account virtual keys (403 / "no persisted bytes available"), so file content fetches need to work end-to-end under scoped non-admin keys. (LIT-3769) - [ ] Streaming file uploads fail with agents SDK in v1.88.0 on /v1/responses API (LIT-3737) - [x] Bedrock Mantle not working for OpenAI (Urgent Bug/Blocker)
## Spend Tracking and Budgets
- [ ] P0: Virtual key spend limits are not being enforced (LIT-3772) - [x] P0: Anthropic cost tracking is inaccurate for cached and tiered usage (LIT-3771)
## UI + Auth
- [ ] Budget reset timing isn't recalculated correctly when the duration changes. (LIT-3362) - [x] Add verification_uri_complete to CLI SSO device flow (LIT-3693) (Urgent Blocking Bug/Regression)
## Performance
- [x] Guardrails con…