[bot] Google GenAI: Image editing API (`models.edit_image()`) not instrumented

[braintrust-bot]

Summary

The Google GenAI image editing API (Models.edit_image() / AsyncModels.edit_image()) is not instrumented. Calls to edit images using Imagen models produce zero Braintrust tracing. This is the image-editing counterpart to Models.generate_images(), which IS instrumented in this repo.

What is missing

Google GenAI Method	Instrumented?
models.generate_content()	Yes
models.generate_content_stream()	Yes
models.embed_content()	Yes
models.generate_images()	Yes
models.edit_image()	No
async_models.edit_image()	No
models.generate_videos()	No (tracked in #236)

The edit_image() method is the SDK entry point for AI-powered image editing operations including inpainting, outpainting, style transfer, and subject-based editing using Imagen models. It accepts a prompt, a reference image, and editing config, and returns an edited image result.

Minimum instrumentation

At minimum, both sync and async edit_image() should create spans capturing:

• Input: prompt text, reference image metadata, model name, editing config
• Output: edited image metadata (format, dimensions)
• Metrics: latency
• Metadata: model, editing parameters

The pattern should mirror the existing ModelsGenerateImagesPatcher / AsyncModelsGenerateImagesPatcher and reuse the _generate_images_wrapper / _async_generate_images_wrapper infrastructure if the response format is compatible, or define new _edit_image_wrapper / _async_edit_image_wrapper functions if not.

Braintrust docs status

not_found — The Gemini integration page documents generate_content, streaming, function calling, structured outputs, thinking tokens, and context caching. Image generation (generate_images) is supported. Image editing (edit_image) is not mentioned.

Upstream sources

• Google GenAI image editing guide: https://ai.google.dev/gemini-api/docs/image-generation (covers both generation and editing)
• Google GenAI Python SDK reference: https://googleapis.github.io/python-genai/ — Models.edit_image() and AsyncModels.edit_image() are public methods
• SDK source: https://github.com/googleapis/python-genai/blob/main/google/genai/models.py — edit_image at line 6627 (sync), line 8687 (async), with internal _edit_image implementations

Local files inspected

• py/src/braintrust/integrations/google_genai/patchers.py — defines ModelsGenerateImagesPatcher and AsyncModelsGenerateImagesPatcher but no edit_image patchers; zero references to edit_image
• py/src/braintrust/integrations/google_genai/tracing.py — contains _generate_images_wrapper and _async_generate_images_wrapper but no edit_image wrappers
• py/src/braintrust/integrations/google_genai/integration.py — integration class registers image generation patchers but no image editing patchers
• py/src/braintrust/integrations/google_genai/test_google_genai.py — no edit_image test cases
• py/pyproject.toml — Google GenAI matrix: latest pinned to google-genai==1.74.0

Relationship to existing issues

• Google GenAI: generate_images() (Imagen) not instrumented #154 (closed) added generate_images instrumentation — edit_image is the adjacent gap on the same Models class
• [bot] Google GenAI: Video generation API (models.generate_videos()) not instrumented #236 tracks generate_videos (another missing generative method on Models)