Is your feature request related to a problem? Please describe.
GoModel does not implement the OpenAI-compatible /v1/images/generations endpoint. Users who want a single OpenAI-compatible gateway for both text and image generation workloads currently cannot route image generation through GoModel at all.
Confirmed missing: no route, handler, or provider adapter exists anywhere in the codebase. There is an existing core.CategoryImage / "image_generation" model-catalog classification (used only to advertise which models support image generation in the models list), but nothing consumes it to actually serve a request.
Describe the solution you'd like
Add a /v1/images/generations endpoint accepting the standard OpenAI request shape (prompt, model, size, quality, n, response format) and returning base64 or URL image data in the standard response shape.
This needs a new provider adapter rather than reusing the existing chat/responses translation helpers, since image generation requests and responses are not text/tool-call shaped like those are.
Describe alternatives you've considered
None specific to scope - the request/response shape is dictated by the existing OpenAI API.
Additional context
Reasonable first-provider targets, since both are already registered provider types in GoModel with existing auth plumbing:
- OpenAI, whose native
/v1/images/generations is close to a direct pass-through.
- Gemini (image generation model family), as a second provider once the OpenAI path is working.
Medium scope: this is a genuinely new endpoint and adapter, not a small tweak to existing code.
Is your feature request related to a problem? Please describe.
GoModel does not implement the OpenAI-compatible
/v1/images/generationsendpoint. Users who want a single OpenAI-compatible gateway for both text and image generation workloads currently cannot route image generation through GoModel at all.Confirmed missing: no route, handler, or provider adapter exists anywhere in the codebase. There is an existing
core.CategoryImage/"image_generation"model-catalog classification (used only to advertise which models support image generation in the models list), but nothing consumes it to actually serve a request.Describe the solution you'd like
Add a
/v1/images/generationsendpoint accepting the standard OpenAI request shape (prompt,model,size,quality,n, response format) and returning base64 or URL image data in the standard response shape.This needs a new provider adapter rather than reusing the existing chat/responses translation helpers, since image generation requests and responses are not text/tool-call shaped like those are.
Describe alternatives you've considered
None specific to scope - the request/response shape is dictated by the existing OpenAI API.
Additional context
Reasonable first-provider targets, since both are already registered provider types in GoModel with existing auth plumbing:
/v1/images/generationsis close to a direct pass-through.Medium scope: this is a genuinely new endpoint and adapter, not a small tweak to existing code.