Search gives your agents read access to the internet, extending their context with the most up‑to‑date information.

Now, Search is natively in our dashboard, allowing you to sample queries before taking them to production.

Stagehand v4 caches act(), observe(), and extract() results on Browserbase's servers. Now you control it.

Pass cache to set a hit-count threshold - how many identical results Browserbase must see before it starts serving that result from the cache. A hit skips inference entirely: no model call, no tokens.

const stagehand = await Stagehand.create({
  browser,
  cache: { threshold: 2 }, // Serve hits after two identical results
});

// Prime this one step aggressively: the second call is a hit
await stagehand.act("click the Sign in button", {
  cache: { threshold: 1 },
});

Read the docs →

Plus:

• Every call can override the instance setting; pass cache: false to skip the cache for a single step.
• Results carry metadata.cache with a HIT / MISS / DISABLED status, a miss reason, and the tokens a hit saved.
• Model configuration stays out of the cache key, so switching models does not invalidate your cache.

You can now assign human-readable names your Contexts for easier recall, and view all previously created Contexts in the dashboard.

Create new Contexts, delete old ones, and see all previous browser sessions created using a specific Context.

Enterprise customers can now deliver Chrome DevTools Protocol (CDP) session logs straight to their own AWS S3 bucket.

BYOS is available on Enterprise plans and enabled per project. See the BYOS setup guide to get started, or contact your account manager.

Playwright was built for testing, we built Stagehand for your agent: improved context management, self-healing actions, and iframe support.

Stagehand now lives in the browser as an extension, bringing the local experience to remote browsers. Being in the browser means lower round trip time, which makes actions execute faster and minimizes potential CDP race conditions.

We benchmarked v4 extensively and found that it’s performs 2x faster than Playwright, while being ~80% more token efficient.

Fetch lets your agents read any page on the internet as Markdown or HTML. You can now view past Fetch calls, inspect their responses, and rerun them directly in the playground.

Picking a model for every browser action is manual work, and the wrong pick costs you either accuracy or money. Model Router does it for you.

Set model: "auto", and Browserbase selects the model for each call server-side, balancing accuracy, speed, and cost automatically.

const stagehand = new Stagehand({
  env: "BROWSERBASE",
  model: "auto",
});

Teams typically see 30–40% lower inference cost than pinning a single model for every call, with no code changes.

Model Router builds on Model Gateway and requires env: "BROWSERBASE" and a Browserbase API key - no provider keys needed.

Read the docs →

One of our most requested features, you can now download MP4 Session Recordings via the Browserbase API and Dashboard.

Read more: https://docs.browserbase.com/platform/browser/observability/recording-downloads

Stagehand 3.7.0 is live.

Use setDomainPolicy() to allowlist or blocklist the domains your agent can visit. Stagehand automatically closes any popup that tries to violate the policy.

await stagehand.context.setDomainPolicy({
  allowedDomains: ["www.browserbase.com"],
});

Read the docs →

Plus:

• New CUA models: google/gemini-3.5-flash and the GPT-5.6 family (sol, luna, terra).

Fixes:

• CUA keypress actions now send key combinations as a single chord.
• Malformed UTF-16 snapshot text is repaired before it reaches model prompts.
• Removed a noisy AI SDK warning logged on every hybrid/DOM agent.execute() call.
• Fixed a crash when passing an MCP Client instance into agent({ integrations }).

Browserbase Agents is a managed solution that enable you to deploy agents on the web with a simple API call. No code to maintain or to deploy, you describe the task goal in natural language and trigger a run using a simple prompt.

Browserbase Agents is the perfect fit to reliably run at scale if you want to:

• Automate a site too dynamic or varied to script reliably
• Scale automation past a handful of sites without one script per target
• Ship browser agents without standing up infrastructure (compute, orchestration, evals)
• Cut maintenance on fragile, high-upkeep scripts

Learn more

Stagehand 3.6.0 is live.

With WebMCP support, a page can expose typed, first-party tools directly to your agent — so Stagehand calls those tools instead of inferring actions from the DOM. Discover what a page offers with page.listWebMCPTools() and run one with page.invokeWebMCPTool().

const page = stagehand.context.pages()[0];
await page.goto("https://example.com");

// Discover the tools the page exposes
const tools = await page.listWebMCPTools();

// Invoke one directly
const result = await page.invokeWebMCPTool(
  "searchFlights",
  { origin: "SFO", destination: "JFK" },
  { frameId: tools[0].frameId },
);

Read the WebMCP docs →

Plus:

• Azure OpenAI auth via Microsoft Entra ID.
• claude-fable-5 support, with adaptive thinking and the new xhigh effort.
• Smarter act caching — URLs that differ only in query-param order now hit the cache.

Fixes:

• Shadow-root XPath resolution so deterministic actions can target elements inside web components.
• Security and dependency updates.

Stagehand 3.5.0 is live.

Pass screenshot: true to extract() to send the viewport screenshot alongside the accessibility tree, so the model can read charts, layouts, and visual-only content the a11y tree misses.

const data = await page.extract({
  instruction: "extract the pricing tiers and prices",
  schema,
  screenshot: true,
});

Plus:

• Native clipboard API for copy/paste flows.
• Claude Opus 4.8 is now supported for Computer Use (CUA).
• Google Vertex AI auth parameters across the SDK and API schemas.
• Use ignoreDefaultArgs to drop chrome-launcher's built-in flags (e.g. --disable-extensions) when running locally.
• Snapshots now include [selected] / [checked] state for more reliable form handling.
• Constructor and per-request model config now forwards to API-backed sessions.

Fixes:

• Anthropic CUA triple_click action mapping.
• structuredOutputMode for newer Anthropic models.
• Local viewport settings are now respected when attaching over CDP.