[BOT ISSUE] OpenAI Images API not instrumented (openai + ruby-openai gems)

[braintrust-bot]

Summary

The OpenAI Images API (client.images.generate) is available in both the official openai gem and the ruby-openai gem, but is not instrumented by this SDK. Image generation (DALL-E, gpt-image-1) is a stable generative AI API that creates images from text prompts. This is the same class of gap as the already-filed audio API issue (#146), and is distinct from the RubyLLM paint() gap (#136) which covers a different library.

What is missing

Official openai gem

Three resource methods under OpenAI::Resources::Images:

• client.images.generate(...) — Text-to-image generation using DALL-E 3 or gpt-image-1. Accepts prompt, model, size, quality, style, n, response_format, background, output_format, output_compression, and more. Returns typed OpenAI::Models::ImagesResponse.
• client.images.edit(...) — Image editing with text prompts. Accepts image file + prompt + mask.
• client.images.create_variation(...) — Image variations from an existing image.

The official SDK also has streaming variants (generate_stream_raw, edit_stream_raw).

ruby-openai gem

Equivalent methods with hash-based parameters:

• client.images.generate(parameters: { prompt:, model:, size:, quality:, ... })
• client.images.edit(parameters: { ... })
• client.images.variations(parameters: { ... })

Documented in the README with DALL-E 2 and DALL-E 3 examples.

Expected instrumentation

A new ImagesPatcher (following the existing ModerationsPatcher pattern) for each gem:

Image generation spans should capture:

• Input: the text prompt
• Metadata: model (e.g., dall-e-3, gpt-image-1), size, quality, style, n, response_format, provider, endpoint
• Output: image URL(s) or metadata (not the binary/base64 image data itself)
• Metrics: revised_prompt (when the model modifies the original prompt)

Braintrust docs status

not_found — Braintrust docs at https://www.braintrust.dev/docs/guides/tracing and the integrations pages do not mention image generation instrumentation. All examples focus on chat completions, embeddings, and audio.

Upstream sources

• Official OpenAI Ruby SDK: https://github.com/openai/openai-ruby — OpenAI::Resources::Images class with generate, edit, create_variation methods
• OpenAI Images API docs: https://platform.openai.com/docs/api-reference/images
• ruby-openai gem: https://github.com/alexrudall/ruby-openai — client.images.generate(parameters: {}) documented in README

Local repo files inspected

• lib/braintrust/contrib/openai/patcher.rb — defines ChatPatcher, ResponsesPatcher, ModerationsPatcher but no ImagesPatcher
• lib/braintrust/contrib/openai/instrumentation/ — contains chat.rb, responses.rb, moderations.rb but no images.rb
• lib/braintrust/contrib/ruby_openai/patcher.rb — defines ChatPatcher, ResponsesPatcher, ModerationsPatcher but no ImagesPatcher
• lib/braintrust/contrib/ruby_openai/instrumentation/ — contains chat.rb, responses.rb, moderations.rb but no images.rb
• Grep for images, dall, ImagesPatcher across lib/braintrust/ returns zero relevant matches