feat: Split google into gemini and vertexai

pyproject.toml

@@ -19,7 +19,7 @@ dependencies = [
 [project.optional-dependencies]
 
 all = [
-  "any-llm-sdk[mistral,anthropic,huggingface,google,cohere,cerebras,fireworks,groq,aws,azure,azureopenai,watsonx,together,sambanova,ollama,moonshot,nebius,xai,databricks,deepseek,inception,openai,openrouter,portkey,lmstudio,llama,voyage,perplexity,llamafile,llamacpp]"
+  "any-llm-sdk[mistral,anthropic,huggingface,gemini,vertexai,cohere,cerebras,fireworks,groq,aws,azure,azureopenai,watsonx,together,sambanova,ollama,moonshot,nebius,xai,databricks,deepseek,inception,openai,openrouter,portkey,lmstudio,llama,voyage,perplexity,llamafile,llamacpp]"
 ]
 
 perplexity = []
@@ -32,7 +32,11 @@ anthropic = [
   "anthropic",
 ]
 
-google = [
+gemini = [
+  "google-genai",
+]
+
+vertexai = [
   "google-genai",
 ]
 

src/any_llm/provider.py

@@ -43,7 +43,7 @@ class ProviderName(StrEnum):
     DATABRICKS = "databricks"
     DEEPSEEK = "deepseek"
     FIREWORKS = "fireworks"
-    GOOGLE = "google"
+    GEMINI = "gemini"
     GROQ = "groq"
     HUGGINGFACE = "huggingface"
     INCEPTION = "inception"
@@ -60,6 +60,7 @@ class ProviderName(StrEnum):
     PORTKEY = "portkey"
     SAMBANOVA = "sambanova"
     TOGETHER = "together"
+    VERTEXAI = "vertexai"
     VOYAGE = "voyage"
     WATSONX = "watsonx"
     XAI = "xai"

src/any_llm/providers/gemini/__init__.py

@@ -0,0 +1,3 @@
+from .gemini import GeminiProvider
+
+__all__ = ["GeminiProvider"]

src/any_llm/providers/google/google.py → src/any_llm/providers/gemini/base.py

@@ -1,10 +1,10 @@
-import os
+from abc import abstractmethod
 from collections.abc import AsyncIterator, Sequence
-from typing import Any
+from typing import TYPE_CHECKING, Any, Literal, cast
 
 from pydantic import BaseModel
 
-from any_llm.exceptions import MissingApiKeyError, UnsupportedParameterError
+from any_llm.exceptions import UnsupportedParameterError
 from any_llm.provider import ClientConfig, Provider
 from any_llm.types.completion import (
     ChatCompletion,
@@ -38,16 +38,14 @@
 except ImportError as e:
     MISSING_PACKAGES_ERROR = e
 
-# From https://ai.google.dev/gemini-api/docs/openai#thinking
+if TYPE_CHECKING:
+    from google import genai
+
 REASONING_EFFORT_TO_THINKING_BUDGETS = {"minimal": 256, "low": 1024, "medium": 8192, "high": 24576}
 
 
 class GoogleProvider(Provider):
-    """Google Provider using the new response conversion utilities."""
-
-    PROVIDER_NAME = "google"
-    PROVIDER_DOCUMENTATION_URL = "https://cloud.google.com/vertex-ai/docs"
-    ENV_API_KEY_NAME = "GOOGLE_API_KEY/GEMINI_API_KEY"
+    """Base Google Provider class with common functionality for Gemini and Vertex AI."""
 
     SUPPORTS_COMPLETION_STREAMING = True
     SUPPORTS_COMPLETION = True
@@ -58,43 +56,17 @@ class GoogleProvider(Provider):
 
     MISSING_PACKAGES_ERROR = MISSING_PACKAGES_ERROR
 
-    def __init__(self, config: ClientConfig) -> None:
-        """Initialize Google GenAI provider."""
-        self._verify_no_missing_packages()
-        self.config = config
-        self.use_vertex_ai = os.getenv("GOOGLE_USE_VERTEX_AI", "false").lower() == "true"
-
-    def _get_client(self, use_vertex_ai: bool, config: ClientConfig) -> "genai.Client":
-        if use_vertex_ai:
-            project_id = os.getenv("GOOGLE_PROJECT_ID")
-            location = os.getenv("GOOGLE_REGION", "us-central1")
-
-            if not project_id:
-                msg = "Google Vertex AI"
-                raise MissingApiKeyError(msg, "GOOGLE_PROJECT_ID")
-
-            return genai.Client(
-                vertexai=True,
-                project=project_id,
-                location=location,
-                **(config.client_args if config.client_args else {}),
-            )
-
-        api_key = getattr(config, "api_key", None) or os.getenv("GEMINI_API_KEY") or os.getenv("GOOGLE_API_KEY")
-
-        if not api_key:
-            msg = "Google Gemini Developer API"
-            raise MissingApiKeyError(msg, "GEMINI_API_KEY/GOOGLE_API_KEY")
-
-        return genai.Client(api_key=api_key, **(config.client_args if config.client_args else {}))
+    @abstractmethod
+    def _get_client(self, config: ClientConfig) -> "genai.Client":
+        """Get the appropriate client for this provider implementation."""
 
     async def aembedding(
         self,
         model: str,
         inputs: str | list[str],
         **kwargs: Any,
     ) -> CreateEmbeddingResponse:
-        client = self._get_client(self.use_vertex_ai, self.config)
+        client = self._get_client(self.config)
         result = await client.aio.models.embed_content(
             model=model,
             contents=inputs,  # type: ignore[arg-type]
@@ -125,15 +97,13 @@ async def acompletion(
 
         if params.reasoning_effort is None:
             kwargs["thinking_config"] = types.ThinkingConfig(include_thoughts=False)
-        # in "auto" mode, we just don't pass a `thinking_config`
         elif params.reasoning_effort != "auto":
             kwargs["thinking_config"] = types.ThinkingConfig(
                 include_thoughts=True, thinking_budget=REASONING_EFFORT_TO_THINKING_BUDGETS[params.reasoning_effort]
             )
 
         stream = bool(params.stream)
         response_format = params.response_format
-        # Build generation config without duplicating keys (e.g., tools)
         base_kwargs = params.model_dump(
             exclude_none=True,
             exclude={
@@ -148,7 +118,6 @@ async def acompletion(
             },
         )
 
-        # Convert max_tokens to max_output_tokens for Google
         if params.max_tokens is not None:
             base_kwargs["max_output_tokens"] = params.max_tokens
 
@@ -162,7 +131,7 @@ async def acompletion(
         if system_instruction:
             generation_config.system_instruction = system_instruction
 
-        client = self._get_client(self.use_vertex_ai, self.config)
+        client = self._get_client(self.config)
         if stream:
             response_stream = await client.aio.models.generate_content_stream(
                 model=params.model_id,
@@ -184,7 +153,6 @@ async def _stream() -> AsyncIterator[ChatCompletionChunk]:
 
         response_dict = _convert_response_to_response_dict(response)
 
-        # Directly construct ChatCompletion
         choices_out: list[Choice] = []
         for i, choice_item in enumerate(response_dict.get("choices", [])):
             message_dict: dict[str, Any] = choice_item.get("message", {})
@@ -211,8 +179,6 @@ async def _stream() -> AsyncIterator[ChatCompletionChunk]:
                 tool_calls=tool_calls,
                 reasoning=Reasoning(content=reasoning_content) if reasoning_content else None,
             )
-            from typing import Literal, cast
-
             choices_out.append(
                 Choice(
                     index=i,
@@ -241,9 +207,7 @@ async def _stream() -> AsyncIterator[ChatCompletionChunk]:
         )
 
     def list_models(self, **kwargs: Any) -> Sequence[Model]:
-        """
-        Fetch available models from the /v1/models endpoint.
-        """
-        client = self._get_client(self.use_vertex_ai, self.config)
+        """Fetch available models from the /v1/models endpoint."""
+        client = self._get_client(self.config)
         models_list = client.models.list(**kwargs)
         return _convert_models_list(models_list)

src/any_llm/providers/gemini/gemini.py

@@ -0,0 +1,26 @@
+import os
+
+from google import genai
+
+from any_llm.exceptions import MissingApiKeyError
+from any_llm.provider import ClientConfig
+
+from .base import GoogleProvider
+
+
+class GeminiProvider(GoogleProvider):
+    """Gemini Provider using the Google GenAI Developer API."""
+
+    PROVIDER_NAME = "gemini"
+    PROVIDER_DOCUMENTATION_URL = "https://ai.google.dev/gemini-api/docs"
+    ENV_API_KEY_NAME = "GEMINI_API_KEY/GOOGLE_API_KEY"
+
+    def _get_client(self, config: ClientConfig) -> "genai.Client":
+        """Get Gemini API client."""
+        api_key = getattr(config, "api_key", None) or os.getenv("GEMINI_API_KEY") or os.getenv("GOOGLE_API_KEY")
+
+        if not api_key:
+            msg = "Google Gemini Developer API"
+            raise MissingApiKeyError(msg, "GEMINI_API_KEY/GOOGLE_API_KEY")
+
+        return genai.Client(api_key=api_key, **(config.client_args if config.client_args else {}))

src/any_llm/providers/google/utils.py → src/any_llm/providers/gemini/utils.py

@@ -26,7 +26,6 @@ def _convert_tool_spec(openai_tools: list[dict[str, Any]]) -> list[types.Tool]:
             continue
 
         function = tool["function"]
-        # Preserve nested schema details such as items/additionalProperties for arrays/objects
         properties: dict[str, dict[str, Any]] = {}
         for param_name, param_info in function["parameters"]["properties"].items():
             prop: dict[str, Any] = {
@@ -35,12 +34,10 @@ def _convert_tool_spec(openai_tools: list[dict[str, Any]]) -> list[types.Tool]:
             }
             if "enum" in param_info:
                 prop["enum"] = param_info["enum"]
-            # Google requires explicit items for arrays
             if "items" in param_info:
                 prop["items"] = param_info["items"]
             if prop.get("type") == "array" and "items" not in prop:
                 prop["items"] = {"type": "string"}
-            # Google tool schema does not accept additionalProperties; drop it
             properties[param_name] = prop
 
         parameters_dict = {
@@ -85,7 +82,7 @@ def _convert_messages(messages: list[dict[str, Any]]) -> tuple[list[types.Conten
             formatted_messages.append(types.Content(role="user", parts=parts))
         elif message["role"] == "assistant":
             if message.get("tool_calls"):
-                tool_call = message["tool_calls"][0]  # Assuming single function call for now
+                tool_call = message["tool_calls"][0]
                 function_call = tool_call["function"]
 
                 parts = [
@@ -202,7 +199,6 @@ def _create_openai_embedding_response_from_google(
         if embedding.values
     ]
 
-    # Google does not provide usage data in the embedding response
     usage = Usage(prompt_tokens=0, total_tokens=0)
 
     return CreateEmbeddingResponse(
@@ -228,10 +224,8 @@ def _create_openai_chunk_from_google_chunk(
 
     for part in candidate.content.parts:
         if part.thought:
-            # This is a thinking/reasoning part
             reasoning_content += part.text or ""
         else:
-            # Regular content part
             content += part.text or ""
 
     delta = ChoiceDelta(
@@ -247,7 +241,7 @@ def _create_openai_chunk_from_google_chunk(
     )
 
     return ChatCompletionChunk(
-        id=f"chatcmpl-{time()}",  # Google doesn't provide an ID in the chunk
+        id=f"chatcmpl-{time()}",
         choices=[choice],
         created=int(time()),
         model=str(response.model_version),
@@ -256,5 +250,4 @@ def _create_openai_chunk_from_google_chunk(
 
 
 def _convert_models_list(models_list: Pager[types.Model]) -> list[Model]:
-    # Google doesn't provide a creation date for models
     return [Model(id=model.name or "Unknown", object="model", created=0, owned_by="google") for model in models_list]

src/any_llm/providers/vertexai/__init__.py

@@ -0,0 +1,3 @@
+from .vertexai import VertexaiProvider
+
+__all__ = ["VertexaiProvider"]

src/any_llm/providers/vertexai/vertexai.py

@@ -0,0 +1,35 @@
+import os
+from typing import TYPE_CHECKING
+
+from any_llm.exceptions import MissingApiKeyError
+from any_llm.provider import ClientConfig
+from any_llm.providers.gemini.base import GoogleProvider
+
+if TYPE_CHECKING:
+    from google import genai
+
+
+class VertexaiProvider(GoogleProvider):
+    """Vertex AI Provider using Google Cloud Vertex AI."""
+
+    PROVIDER_NAME = "vertexai"
+    PROVIDER_DOCUMENTATION_URL = "https://cloud.google.com/vertex-ai/docs"
+    ENV_API_KEY_NAME = "GOOGLE_PROJECT_ID"
+
+    def _get_client(self, config: ClientConfig) -> "genai.Client":
+        """Get Vertex AI client."""
+        from google import genai
+
+        project_id = os.getenv("GOOGLE_PROJECT_ID")
+        location = os.getenv("GOOGLE_REGION", "us-central1")
+
+        if not project_id:
+            msg = "Google Vertex AI"
+            raise MissingApiKeyError(msg, "GOOGLE_PROJECT_ID")
+
+        return genai.Client(
+            vertexai=True,
+            project=project_id,
+            location=location,
+            **(config.client_args if config.client_args else {}),
+        )

tests/conftest.py

@@ -10,7 +10,8 @@ def provider_reasoning_model_map() -> dict[ProviderName, str]:
     return {
         ProviderName.ANTHROPIC: "claude-sonnet-4-20250514",
         ProviderName.MISTRAL: "magistral-small-latest",
-        ProviderName.GOOGLE: "gemini-2.5-flash",
+        ProviderName.GEMINI: "gemini-2.5-flash",
+        ProviderName.VERTEXAI: "gemini-2.5-flash",
         ProviderName.GROQ: "openai/gpt-oss-20b",
         ProviderName.FIREWORKS: "accounts/fireworks/models/deepseek-r1",
         ProviderName.OPENAI: "gpt-5-nano",
@@ -33,7 +34,8 @@ def provider_model_map() -> dict[ProviderName, str]:
         ProviderName.DATABRICKS: "databricks-meta-llama-3-1-8b-instruct",
         ProviderName.DEEPSEEK: "deepseek-chat",
         ProviderName.OPENAI: "gpt-5-nano",
-        ProviderName.GOOGLE: "gemini-2.5-flash",
+        ProviderName.GEMINI: "gemini-2.5-flash",
+        ProviderName.VERTEXAI: "gemini-2.5-flash",
         ProviderName.MOONSHOT: "moonshot-v1-8k",
         ProviderName.SAMBANOVA: "Meta-Llama-3.1-8B-Instruct",
         ProviderName.TOGETHER: "meta-llama/Llama-3.3-70B-Instruct-Turbo-Free",
@@ -73,7 +75,8 @@ def embedding_provider_model_map() -> dict[ProviderName, str]:
         ProviderName.OLLAMA: "gpt-oss:20b",
         ProviderName.LLAMAFILE: "N/A",
         ProviderName.LMSTUDIO: "text-embedding-nomic-embed-text-v1.5",
-        ProviderName.GOOGLE: "gemini-embedding-001",
+        ProviderName.GEMINI: "gemini-embedding-001",
+        ProviderName.VERTEXAI: "gemini-embedding-001",
         ProviderName.AZURE: "openai/text-embedding-3-small",
         ProviderName.AZUREOPENAI: "azure/<your_deployment_name>",
         ProviderName.VOYAGE: "voyage-3.5-lite",

tests/integration/test_embedding.py

@@ -38,6 +38,6 @@ async def test_embedding_providers_async(
     assert len(result.data) > 0
     for entry in result.data:
         assert all(isinstance(v, float) for v in entry.embedding)
-    if provider not in (ProviderName.GOOGLE, ProviderName.LMSTUDIO):
+    if provider not in (ProviderName.GEMINI, ProviderName.VERTEXAI, ProviderName.LMSTUDIO):
         assert result.usage.prompt_tokens > 0
         assert result.usage.total_tokens > 0

tests/integration/test_reasoning.py

@@ -25,7 +25,7 @@ async def test_completion_reasoning(
 
     model_id = provider_reasoning_model_map[provider]
     extra_kwargs = provider_extra_kwargs_map.get(provider, {})
-    if provider in (ProviderName.ANTHROPIC, ProviderName.GOOGLE, ProviderName.OLLAMA):
+    if provider in (ProviderName.ANTHROPIC, ProviderName.GEMINI, ProviderName.VERTEXAI, ProviderName.OLLAMA):
         extra_kwargs["reasoning_effort"] = "low"
 
     try:
@@ -62,7 +62,7 @@ async def test_completion_reasoning_streaming(
 
     model_id = provider_reasoning_model_map[provider]
     extra_kwargs = provider_extra_kwargs_map.get(provider, {})
-    if provider in (ProviderName.ANTHROPIC, ProviderName.GOOGLE, ProviderName.OLLAMA):
+    if provider in (ProviderName.ANTHROPIC, ProviderName.GEMINI, ProviderName.VERTEXAI, ProviderName.OLLAMA):
         extra_kwargs["reasoning_effort"] = "low"
 
     try: