Support for new Gemini Computer Use Models (#213)

* cheetah updates

* model updates

* changeset

* update viewport

Author: miguelg719

.changeset/horned-giga-cockatoo.md

@@ -0,0 +1,5 @@
+---
+"stagehand": patch
+---
+
+Added support for Gemini Computer Use models

examples/agent_example.py

@@ -43,7 +43,7 @@ async def main():
         self_heal=True,
         system_prompt="You are a browser automation assistant that helps users navigate websites effectively.",
         model_client_options={"apiKey": os.getenv("MODEL_API_KEY")},
-        verbose=1,
+        verbose=2,
     )
 
     # Create a Stagehand client using the configuration object.
@@ -64,9 +64,9 @@ async def main():
 
     console.print("\n▶️ [highlight] Using Agent to perform a task[/]: playing a game of 2048")
     agent = stagehand.agent(
-        model="computer-use-preview",
+        model="gemini-2.5-computer-use-preview-10-2025",
         instructions="You are a helpful web navigation assistant that helps users find information. You are currently on the following page: google.com. Do not ask follow up questions, the user will trust your judgement.",
-        options={"apiKey": os.getenv("MODEL_API_KEY")}
+        options={"apiKey": os.getenv("GEMINI_API_KEY")}
     )
     agent_result = await agent.execute(
         instruction="Play a game of 2048",

pyproject.toml

@@ -9,7 +9,7 @@ description = "Python SDK for Stagehand"
 readme = "README.md"
 classifiers = [ "Programming Language :: Python :: 3", "License :: OSI Approved :: MIT License", "Operating System :: OS Independent",]
 requires-python = ">=3.9"
-dependencies = [ "httpx>=0.24.0", "python-dotenv>=1.0.0", "pydantic>=1.10.0", "playwright>=1.42.1", "requests>=2.31.0", "browserbase>=1.4.0", "rich>=13.7.0", "openai>=1.99.6", "anthropic>=0.51.0", "litellm>=1.72.0,<1.75.0", "nest-asyncio>=1.6.0",]
+dependencies = [ "httpx>=0.24.0", "python-dotenv>=1.0.0", "pydantic>=1.10.0", "playwright>=1.42.1", "requests>=2.31.0", "browserbase>=1.4.0", "rich>=13.7.0", "openai>=1.99.6", "anthropic>=0.51.0", "litellm>=1.72.0,<1.75.0", "nest-asyncio>=1.6.0", "google-genai>=1.40.0"]
 [[project.authors]]
 name = "Browserbase, Inc."
 email = "support@browserbase.com"

stagehand/agent/agent.py

@@ -13,19 +13,24 @@
 )
 from .anthropic_cua import AnthropicCUAClient
 from .client import AgentClient
+from .google_cua import GoogleCUAClient
 from .openai_cua import OpenAICUAClient
 
 MODEL_TO_CLIENT_CLASS_MAP: dict[str, type[AgentClient]] = {
-    "computer-use-preview": OpenAICUAClient,
+    "computer-use-preview-03-11": OpenAICUAClient,
     "claude-3-5-sonnet-latest": AnthropicCUAClient,
     "claude-3-7-sonnet-latest": AnthropicCUAClient,
     "claude-sonnet-4-20250514": AnthropicCUAClient,
+    "claude-sonnet-4-5-20250929": AnthropicCUAClient,
+    "gemini-2.5-computer-use-preview-10-2025": GoogleCUAClient,
 }
 MODEL_TO_PROVIDER_MAP: dict[str, AgentProvider] = {
-    "computer-use-preview": AgentProvider.OPENAI,
+    "computer-use-preview-03-11": AgentProvider.OPENAI,
     "claude-3-5-sonnet-20240620": AgentProvider.ANTHROPIC,
     "claude-3-7-sonnet-20250219": AgentProvider.ANTHROPIC,
     "claude-sonnet-4-20250514": AgentProvider.ANTHROPIC,
+    "claude-sonnet-4-5-20250929": AgentProvider.ANTHROPIC,
+    "gemini-2.5-computer-use-preview-10-2025": AgentProvider.GOOGLE,
     # Add more mappings as needed
 }
 

stagehand/agent/anthropic_cua.py

@@ -62,7 +62,7 @@ def __init__(
         )
 
         dimensions = (
-            (viewport["width"], viewport["height"]) if viewport else (1024, 768)
+            (viewport["width"], viewport["height"]) if viewport else (1288, 711)
         )  # Default dimensions
         if self.config:
             if hasattr(self.config, "display_width") and self.config.display_width is not None:  # type: ignore