fix(vertex_ai): pass through native Gemini imageConfig params for image generation

[Chesars]

Relevant issues

Fixes #21070

Pre-Submission checklist

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• 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

Type

🐛 Bug Fix

Changes

aspectRatio and imageSize parameters were dropped when calling Vertex AI Gemini image generation models (gemini-3-pro-image-preview, gemini-2.5-flash-image).

Fix:

• Added aspectRatio, aspect_ratio, imageSize, image_size to get_supported_openai_params() so they pass validation
• Added explicit mapping in map_openai_params() to normalize snake_case variants to camelCase
• 5 new unit tests covering both camelCase and snake_case variants

*Example

response = litellm.image_generation(
    model="vertex_ai/gemini-3-pro-image-preview",
    prompt="A landscape photo of mountains",
    aspectRatio="9:16",
    imageSize="4K",
)

After this fix, both params are correctly forwarded to the Gemini API

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Feb 19, 2026 6:42pm

[greptile-apps]

Greptile Summary

Fixed silently dropped aspectRatio and imageSize parameters when calling Vertex AI Gemini image generation models (gemini-3-pro-image-preview, gemini-2.5-flash-image).

Key changes:

• Added aspectRatio, aspect_ratio, imageSize, image_size to supported parameters in get_supported_openai_params()
• Added explicit mapping in map_openai_params() to normalize snake_case variants to camelCase
• 5 new unit tests covering both camelCase and snake_case variants (all mock-only, no network calls)

The fix ensures these native Gemini parameters flow correctly through the validation and transformation pipeline to reach the Gemini API inside generationConfig.imageConfig.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• The changes are isolated to Vertex AI Gemini image generation, add well-tested parameter mapping for native Gemini params, follow existing code patterns perfectly, include comprehensive unit tests with no network calls, and fix a genuine bug where parameters were silently dropped
• No files require special attention

Important Files Changed

Filename	Overview
litellm/llms/vertex_ai/image_generation/vertex_gemini_transformation.py	Added support for aspectRatio and imageSize parameters in both camelCase and snake_case variants, properly mapping them through validation and transformation
tests/test_litellm/llms/vertex_ai/image_generation/test_vertex_ai_image_generation_transformation.py	Added 5 comprehensive unit tests covering both camelCase and snake_case variants of new parameters, all tests are mock-only with no network calls

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[User calls litellm.image_generation with aspectRatio/imageSize] --> B[get_optional_params_image_gen]
    B --> C[Provider config detected: VertexAIGeminiImageGenerationConfig]
    C --> D[get_supported_openai_params validates parameters]
    D --> E{Parameter variant?}
    E -->|snake_case| F[map_openai_params normalizes to camelCase]
    E -->|camelCase| F
    F --> G[transform_image_generation_request]
    G --> H[Builds imageConfig in generationConfig]
    H --> I[Parameters sent to Vertex AI Gemini API]

Loading

_{Last reviewed commit: 155bed5}

[greptile-apps]

_{2 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}