Litellm oss staging 030626

[Sameerlite]

Relevant issues

Linear ticket

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have added meaningful tests
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible; it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

CI (LiteLLM team)

CI status guideline:

• 50-55 passing tests: main is stable with minor issues.
• 45-49 passing tests: acceptable but needs attention
• <= 40 passing tests: unstable; be careful with your merges and assess the risk.

• Branch creation CI run
  Link:

• CI run for the last commit
  Link:

• Merge / cherry-pick CI run
  Links:

Screenshots / Proof of Fix

Type

🆕 New Feature
🐛 Bug Fix
🧹 Refactoring
📖 Documentation
🚄 Infrastructure
✅ Test

Changes

[codecov]

Codecov Report

❌ Patch coverage is 94.44444% with 15 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
litellm/llms/gemini/common_utils.py	88.54%	11 Missing ⚠️
litellm/litellm_core_utils/llm_cost_calc/utils.py	88.88%	2 Missing ⚠️
...llm/llms/gemini/image_generation/transformation.py	85.71%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
13 out of 14 committers have signed the CLA.

✅ IshaMeera
✅ DmitriyAlergant
✅ suleimanelkhoury
✅ josx
✅ Thomas-Mildner
✅ Sameerlite
✅ yanismiraoui
✅ Cyberfilo
✅ emerzon
✅ mateo-berri
✅ zzw-math
✅ aneeshsangvikar
✅ vrlo
❌ 1qh
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[greptile-apps]

Greptile Summary

This staging PR bundles several independent fixes and additions: a new Inception (Mercury) provider integration, a Gemini image generation/edit refactor with shared utilities, auth budget-enforcement restructuring, and a handful of targeted bug fixes across Gemini/Vertex message conversion, OTel span propagation, and the responses-bridge handler.

• Inception provider: Adds InceptionChatConfig (OpenAI-compatible chat) and InceptionTextCompletionConfig (FIM at /v1/fim/completions) following the established OpenAILikeChatConfig pattern, with comprehensive mock-based tests covering endpoint routing, key-leak prevention, and parameter mapping.
• Auth refactor: Moves _check_end_user_budget out of get_end_user_object into common_checks() (regular path) and _run_post_custom_auth_checks (custom-auth path), ensuring budget enforcement happens after skip-budget-checks logic can be evaluated. Tests are updated to reflect the new contract.
• Gemini image refactor: Extracts shared aspect-ratio/image-size mapping and ImageUsage transformation into dedicated modules consumed by both image-edit and image-generation transformations; supports_gemini_image_size now queries get_model_info first before falling back to model-name heuristics.

Confidence Score: 5/5

Safe to merge — budget enforcement is fully covered on all auth paths, the new Inception integration is isolated behind a new provider identifier, and all changed behaviour is backed by focused mock tests.

The auth restructuring is consistent end-to-end: common_checks() retains the _check_end_user_budget call at line 738 for the standard path, and _run_post_custom_auth_checks enforces it for the custom-auth path. The Gemini and Vertex fixes are narrowly scoped with dedicated regression tests. The Inception provider adds no new risk to existing flows.

No files require special attention.

Important Files Changed

Filename	Overview
litellm/llms/inception/chat/transformation.py	New OpenAI-compatible Inception chat provider following the established OpenAILikeChatConfig pattern with proper key-resolution and API-base security logic.
litellm/proxy/auth/auth_checks.py	Removes _check_end_user_budget from get_end_user_object so budget enforcement happens in common_checks() instead; _end_user_id_exists_in_db no longer re-raises BudgetExceededError which is consistent with the new contract.
litellm/proxy/auth/user_api_key_auth.py	Adds explicit _check_end_user_budget call inside _run_post_custom_auth_checks for the custom-auth path when custom_auth_run_common_checks is unset, closing the budget-enforcement gap.
litellm/llms/gemini/common_utils.py	Adds shared image-generation utilities consumed by both image-edit and image-generation transformations.
litellm/llms/vertex_ai/gemini/transformation.py	Fixes last_message_with_tool_calls being overwritten by a text-only assistant message.
litellm/router.py	Adds the missing elif custom_model_info branch so models with only custom pricing receive their ModelInfo correctly.

_{Reviews (7): Last reviewed commit: "Merge branch 'litellm_internal_staging' ..." | Re-trigger Greptile}

[Sameerlite]

@greptileai

[Sameerlite]

@greptileai

[Sameerlite]

@greptileai

[veria-ai]

PR overview

All previously flagged issues have been addressed. No open security concerns remain on this pull request.

Security review

No open security issues remain on this pull request.

Fixed/addressed: 2 · PR risk: 0/10

[mateo-berri]

@greptileai

[mateo-berri]

@greptileai

[mateo-berri]

@greptileai

[mateo-berri]

LGTM; thanks!