@spec drop_openai_only_options(keyword(), String.t()) :: {keyword(), [String.t()]}

Drops the options only the OpenAI Images wire format has fields for.

:background, :moderation, :output_compression, and :input_fidelity are part of this shared schema, so every provider's image path accepts them. A provider whose endpoint has no such field must drop them in its ReqLLM.Provider.translate_options/3 image clause: an option that survives translation reaches the Req pipeline unregistered and raises there, instead of surfacing through :on_unsupported like any other lossy translation.

Returns {opts, warnings}, with provider_label naming the provider in each warning.

@spec generate_image(
  ReqLLM.model_input(),
  String.t() | list() | ReqLLM.Context.t(),
  keyword()
) ::
  {:ok, ReqLLM.Response.t()} | {:error, term()}

Generates images using an AI model with full response metadata.

Returns a canonical ReqLLM.Response where images are represented as message content parts.

@spec schema() :: NimbleOptions.t()

Returns the base image generation options schema.

@spec stream_image(
  ReqLLM.model_input(),
  String.t() | list() | ReqLLM.Context.t(),
  keyword()
) ::
  {:ok, ReqLLM.StreamResponse.t()} | {:error, term()}

Streams image generation, yielding preview frames before the final image.

See ReqLLM.stream_image/3 for the chunk contract. Supported for OpenAI and Azure gpt-image models; other providers return ReqLLM.Error.Invalid.Parameter.

@spec stream_image!(
  ReqLLM.model_input(),
  String.t() | list() | ReqLLM.Context.t(),
  keyword()
) ::
  ReqLLM.StreamResponse.t() | no_return()

Streams image generation, raising on error.

Same as stream_image/3 but raises the error struct instead of returning {:error, error}. Only errors raised before the request starts are surfaced here; failures mid-stream still surface while enumerating or through ReqLLM.StreamResponse.to_response/1.

@spec supported_models() :: [String.t()]

Returns a list of model specs that likely support image generation.

Uses capability metadata when present, otherwise falls back to a conservative name-based heuristic (models containing "image" or "imagen").

@spec validate_model(ReqLLM.model_input()) ::
  {:ok, LLMDB.Model.t()} | {:error, term()}

Validates that a model supports image generation operations.