Support for new Gemini Computer Use Models (#1110)

# why
Adding support for Gemini's new Computer Use model

# what changed
We partnered with Google Deepmind to help integrate and test their new
Computer Use models.

<img width="1238" height="655" alt="Screenshot 2025-10-07 at 1 14 44 PM"
src="https://github.com/user-attachments/assets/af0d854a-8e55-4937-a071-10335497f686"
/>

The new model tag `gemini-2.5-pro-computer-use-preview-10-2025` is
available for Stagehand Agent. You can try it today with the example
`cua-example.ts`

To learn more, check out the blog post
[https://www.browserbase.com/blog/evaluating-browser-agents](https://www.browserbase.com/blog/evaluating-browser-agents)

---------

Co-authored-by: tkattkat <[email protected]>
Co-authored-by: Kylejeong2 <[email protected]>
Co-authored-by: Sameel <[email protected]>

Author: 4 people

.changeset/wicked-ducks-share.md

@@ -0,0 +1,5 @@
+---
+"@browserbasehq/stagehand": patch
+---
+
+Add support for new Gemini Computer Use models

examples/cua-example.ts

@@ -26,14 +26,15 @@ async function main() {
 
     // Create a computer use agent
     const agent = stagehand.agent({
-      provider: "openai",
-      // For Anthropic, use claude-sonnet-4-20250514 or claude-3-7-sonnet-latest
-      model: "computer-use-preview",
+      provider: "google",
+      // For Anthropic, use claude-sonnet-4-20250514 or claude-sonnet-4-5-20250929
+      // For OpenAI, use computer-use-preview-03-11
+      model: "gemini-2.5-computer-use-preview-10-2025",
       instructions: `You are a helpful assistant that can use a web browser.
       You are currently on the following page: ${page.url()}.
       Do not ask follow up questions, the user will trust your judgement.`,
       options: {
-        apiKey: process.env.OPENAI_API_KEY,
+        apiKey: process.env.GOOGLE_API_KEY,
       },
     });
 

lib/agent/AgentProvider.ts

@@ -8,6 +8,7 @@ import { ToolSet } from "ai/dist";
 import { AgentClient } from "./AgentClient";
 import { AnthropicCUAClient } from "./AnthropicCUAClient";
 import { OpenAICUAClient } from "./OpenAICUAClient";
+import { GoogleCUAClient } from "./GoogleCUAClient";
 
 // Map model names to their provider types
 export const modelToAgentProviderMap: Record<string, AgentType> = {
@@ -16,6 +17,7 @@ export const modelToAgentProviderMap: Record<string, AgentType> = {
   "claude-3-7-sonnet-latest": "anthropic",
   "claude-sonnet-4-20250514": "anthropic",
   "claude-sonnet-4-5-20250929": "anthropic",
+  "gemini-2.5-computer-use-preview-10-2025": "google",
 };
 
 /**
@@ -64,9 +66,16 @@ export class AgentProvider {
             clientOptions,
             tools,
           );
+        case "google":
+          return new GoogleCUAClient(
+            type,
+            modelName,
+            userProvidedInstructions,
+            clientOptions,
+          );
         default:
           throw new UnsupportedModelProviderError(
-            ["openai", "anthropic"],
+            ["openai", "anthropic", "google"],
             "Computer Use Agent",
           );
       }

lib/agent/AnthropicCUAClient.ts

@@ -29,7 +29,7 @@ export class AnthropicCUAClient extends AgentClient {
   private baseURL?: string;
   private client: Anthropic;
   public lastMessageId?: string;
-  private currentViewport = { width: 1024, height: 768 };
+  private currentViewport = { width: 1288, height: 711 };
   private currentUrl?: string;
   private screenshotProvider?: () => Promise<string>;
   private actionHandler?: (action: AgentAction) => Promise<void>;
@@ -290,7 +290,7 @@ export class AnthropicCUAClient extends AgentClient {
 
           logger({
             category: "agent",
-            message: `Found text block: ${textBlock.text.substring(0, 50)}...`,
+            message: `Found text block: ${textBlock.text}`,
             level: 2,
           });
         } else {