fix Gemini image base and provider docs

nanobot/providers/image_generation.py

@@ -129,6 +129,11 @@ _IMAGE_GEN_PROVIDERS: dict[str, type[ImageGenerationProvider]] = {}
 
 
 def register_image_gen_provider(cls: type[ImageGenerationProvider]) -> None:
+    """Register an image provider at import time only.
+
+    The registry is populated by module side effects so provider discovery
+    stays lazy and consistent across the process.
+    """
     name = cls.provider_name
     if not name:
         raise ValueError(f"{cls.__name__} must set provider_name")
@@ -204,7 +209,7 @@ class ImageGenerationProvider(ABC):
         image_size: str | None = None,
     ) -> GeneratedImageResponse: ...
 
-    def _require_images(self, images: list[str], data: dict[str, Any]) -> None:
+    def _ensure_images(self, images: list[str], data: dict[str, Any]) -> None:
         if images:
             return
         provider_error = data.get("error") if isinstance(data, dict) else None
@@ -311,7 +316,7 @@ class OpenRouterImageGenerationClient(ImageGenerationProvider):
                 if isinstance(url_value, str) and url_value.startswith("data:image/"):
                     images.append(url_value)
 
-        self._require_images(images, data)
+        self._ensure_images(images, data)
 
         return GeneratedImageResponse(
             images=images,
@@ -413,7 +418,7 @@ class AIHubMixImageGenerationClient(ImageGenerationProvider):
         payload = response.json()
         images = await _aihubmix_images_from_payload(client, payload)
 
-        self._require_images(images, payload)
+        self._ensure_images(images, payload)
 
         return GeneratedImageResponse(images=images, content="", raw=payload)
 
@@ -446,9 +451,9 @@ class GeminiImageGenerationClient(ImageGenerationProvider):
         return "https://generativelanguage.googleapis.com/v1beta"
 
     def _resolve_base_url(self, api_base: str | None) -> str:
-        # The Gemini provider's registry default_api_base is the OpenAI-compat
+        # Gemini chat completions use the registry's OpenAI-compatible shim.
-        # shim (.../v1beta/openai/), which has no image endpoints.
+        # Image generation must hit the native Generative Language API, so we
-        # Skip the registry lookup and use the native API base directly.
+        # intentionally bypass the shared registry lookup here.
         if api_base:
             return api_base.rstrip("/")
         return self._default_base_url()
@@ -522,7 +527,7 @@ class GeminiImageGenerationClient(ImageGenerationProvider):
             if isinstance(b64, str) and b64:
                 images.append(f"data:{mime};base64,{b64}")
 
-        self._require_images(images, data)
+        self._ensure_images(images, data)
 
         return GeneratedImageResponse(images=images, content="", raw=data)
 
@@ -580,7 +585,7 @@ class GeminiImageGenerationClient(ImageGenerationProvider):
                     if b64:
                         images.append(f"data:{mime};base64,{b64}")
 
-        self._require_images(images, data)
+        self._ensure_images(images, data)
 
         return GeneratedImageResponse(
             images=images,
@@ -734,7 +739,7 @@ class MiniMaxImageGenerationClient(ImageGenerationProvider):
         payload = response.json()
         images = _minimax_images_from_payload(payload)
 
-        self._require_images(images, payload)
+        self._ensure_images(images, payload)
 
         return GeneratedImageResponse(images=images, content="", raw=payload)
 
@@ -840,7 +845,7 @@ class StepFunImageGenerationClient(ImageGenerationProvider):
         payload = response.json()
         images = _stepfun_images_from_payload(payload)
 
-        self._require_images(images, payload)
+        self._ensure_images(images, payload)
 
         return GeneratedImageResponse(images=images, content="", raw=payload)
 

tests/providers/test_image_generation.py

@@ -348,6 +348,11 @@ async def test_gemini_requires_api_key() -> None:
         await client.generate(prompt="draw", model="imagen-4.0-generate-001")
 
 
+def test_gemini_image_client_uses_native_api_base_by_default() -> None:
+    client = GeminiImageGenerationClient(api_key="AIza-test")
+    assert client.api_base == "https://generativelanguage.googleapis.com/v1beta"
+
+
 @pytest.mark.asyncio
 async def test_gemini_no_images_raises() -> None:
     fake = FakeClient(FakeResponse({"candidates": [{"content": {"parts": [{"text": "sorry"}]}}]}))

tests/providers/test_litellm_kwargs.py

@@ -449,6 +449,12 @@ def test_gemma_routes_to_gemini_provider() -> None:
     assert "gemma" in spec.keywords
 
 
+def test_gemini_spec_keeps_openai_compat_base() -> None:
+    spec = find_by_name("gemini")
+    assert spec is not None
+    assert spec.default_api_base == "https://generativelanguage.googleapis.com/v1beta/openai/"
+
+
 async def test_openrouter_sets_default_attribution_headers() -> None:
     spec = find_by_name("openrouter")
     with patch("nanobot.providers.openai_compat_provider.AsyncOpenAI") as mock_client_cls: