fix(images): restrict webp for gpt-image-2 and add missing model

[Deepakhub202]

Changes being requested

Fix incorrect output_format typing for image generation.

Instead of removing webp globally, this change restricts webp only for gpt-image-2 using conditional types.

• gpt-image-2 → allows only png | jpeg
• other models → still allow webp

Also updates related types to work with the new union-based structure.

Additional context & links

Related issue: #1850

The API ignores output_format: "webp" for gpt-image-2 and always returns PNG, so the SDK should reflect this behavior accurately.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2c729c04ed

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Restore webp in image-generate output_format typing

This change narrows ImageGenerateParamsBase.output_format to png|jpeg for all models, which blocks valid requests for models that still support WebP (for example, the same file still documents WebP in generation responses/events and allows it for edits). In practice, TypeScript users can no longer pass output_format: 'webp' even when using non-gpt-image-2 models, so the SDK type now under-represents supported API behavior; this should be modeled conditionally by model instead of removing webp globally.

Useful? React with 👍 / 👎.

[Deepakhub202]

Hi, I’ve updated the fix to use conditional typing so that webp is only restricted for gpt-image-2 while remaining available for other models. Let me know if any further changes are needed. @pedrobarretocw