Skip to content

Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.

chore(llm): recommend gemini-3-flash-preview in examples and tests

Saurav Pandab00c41b66bf6
diff --git a/.github/ISSUE_TEMPLATE/2_bug_report.yml b/.github/ISSUE_TEMPLATE/2_bug_report.yml
index 0a0845ac3..8d64b3631 100644
--- a/.github/ISSUE_TEMPLATE/2_bug_report.yml
+++ b/.github/ISSUE_TEMPLATE/2_bug_report.yml
@@ -57,7 +57,7 @@ body:
attributes:
label: LLM Model
description: Which LLM model are you using? (Optional)
- placeholder: "e.g. ChatBrowserUse, gpt-4.1-mini, gemini-flash-latest, etc."
+ placeholder: "e.g. ChatBrowserUse, gpt-4.1-mini, gemini-3-flash-preview, etc."
- type: input
id: os
diff --git a/AGENTS.md b/AGENTS.md
index 1d71f5d2e..d4f1fcd72 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -126,7 +126,7 @@ See [Supported Models](https://docs.browser-use.com/supported-models#supported-m
load_dotenv()
async def main():
- llm = ChatGoogle(model="gemini-flash-latest")
+ llm = ChatGoogle(model="gemini-3-flash-preview")
task = "Find the number 1 post on Show HN"
agent = Agent(task=task, llm=llm)
await agent.run()
diff --git a/CLOUD.md b/CLOUD.md
index 9e8ee0487..2456d3e53 100644
--- a/CLOUD.md
+++ b/CLOUD.md
@@ -383,6 +383,8 @@ components:
- value: gemini-2.5-pro
- value: gemini-flash-latest
- value: gemini-flash-lite-latest
+ - value: gemini-3-flash-preview
+ - value: gemini-3-flash-preview-lite
- value: claude-sonnet-4-20250514
- value: gpt-4o
- value: gpt-4o-mini
diff --git a/browser_use/llm/google/chat.py b/browser_use/llm/google/chat.py
index 0356e1022..4ec3a5a36 100644
--- a/browser_use/llm/google/chat.py
+++ b/browser_use/llm/google/chat.py
@@ -35,6 +35,7 @@ VerifiedGeminiModels = Literal[
'gemini-2.5-pro',
'gemini-3-pro-preview',
'gemini-3-flash-preview',
+ 'gemini-3-flash-preview-lite',
'gemma-3-27b-it',
'gemma-3-4b',
'gemma-3-12b',
diff --git a/browser_use/tokens/mappings.py b/browser_use/tokens/mappings.py
index 01a5970b1..d6ef75607 100644
--- a/browser_use/tokens/mappings.py
+++ b/browser_use/tokens/mappings.py
@@ -1,4 +1,6 @@
# Mapping from model_name to LiteLLM model name
MODEL_TO_LITELLM: dict[str, str] = {
'gemini-flash-latest': 'gemini/gemini-flash-latest',
+ 'gemini-3-flash-preview': 'gemini/gemini-3-flash-preview',
+ 'gemini-3-flash-preview-lite': 'gemini/gemini-3-flash-preview-lite',
}
diff --git a/examples/browser/real_browser.py b/examples/browser/real_browser.py
index 0405dc430..f761abe9e 100644
--- a/examples/browser/real_browser.py
+++ b/examples/browser/real_browser.py
@@ -37,7 +37,7 @@ async def main():
browser = Browser.from_system_chrome(profile_directory=profile)
agent = Agent(
- llm=ChatGoogle(model='gemini-flash-latest'),
+ llm=ChatGoogle(model='gemini-3-flash-preview'),
task='go to amazon.com and search for pens to draw on whiteboards',
browser=browser,
)
diff --git a/examples/features/stop_externally.py b/examples/features/stop_externally.py
index ee12e83d9..b750c7fba 100644
--- a/examples/features/stop_externally.py
+++ b/examples/features/stop_externally.py
@@ -13,7 +13,7 @@ load_dotenv()
from browser_use import Agent
-llm = ChatGoogle(model='gemini-flash-latest', temperature=1.0)
+llm = ChatGoogle(model='gemini-3-flash-preview', temperature=1.0)
def check_is_task_stopped():
diff --git a/examples/getting_started/05_fast_agent.py b/examples/getting_started/05_fast_agent.py
index c089c1faa..8923406dc 100644
--- a/examples/getting_started/05_fast_agent.py
+++ b/examples/getting_started/05_fast_agent.py
@@ -31,7 +31,7 @@ async def main():
)
# from browser_use import ChatGoogle
- # llm = ChatGoogle(model='gemini-flash-lite-latest')
+ # llm = ChatGoogle(model='gemini-3-flash-preview-lite')
# 2. Create speed-optimized browser profile
browser_profile = BrowserProfile(
diff --git a/examples/models/gemini.py b/examples/models/gemini.py
index 06b033f0a..d9c5ca053 100644
--- a/examples/models/gemini.py
+++ b/examples/models/gemini.py
@@ -16,7 +16,7 @@ if not api_key:
async def run_search():
- llm = ChatGoogle(model='gemini-flash-latest', api_key=api_key)
+ llm = ChatGoogle(model='gemini-3-flash-preview', api_key=api_key)
agent = Agent(
llm=llm,
task='How many stars does the browser-use repo have?',
diff --git a/skills/cloud/references/api-v2.md b/skills/cloud/references/api-v2.md
index 5be9d2c1d..f6c8f7f0d 100644
--- a/skills/cloud/references/api-v2.md
+++ b/skills/cloud/references/api-v2.md
@@ -302,7 +302,7 @@ Response includes: `{ items: [...], totalItems, pageNumber, pageSize }`
| SessionStatus | `active`, `stopped` |
| BrowserSessionStatus | `active`, `stopped` |
| ProxyCountryCode | `us`, `uk`, `fr`, `it`, `jp`, `au`, `de`, `fi`, `ca`, `in` (+185 more) |
-| SupportedLLMs | `browser-use-llm`, `gpt-4.1`, `gpt-4.1-mini`, `o4-mini`, `o3`, `gemini-2.5-flash`, `gemini-2.5-pro`, `gemini-flash-latest`, `gemini-flash-lite-latest`, `claude-sonnet-4-20250514`, `gpt-4o`, `gpt-4o-mini`, `llama-4-maverick-17b-128e-instruct`, `claude-3-7-sonnet-20250219` |
+| SupportedLLMs | `browser-use-llm`, `gpt-4.1`, `gpt-4.1-mini`, `o4-mini`, `o3`, `gemini-2.5-flash`, `gemini-2.5-pro`, `gemini-flash-latest`, `gemini-flash-lite-latest`, `gemini-3-flash-preview`, `gemini-3-flash-preview-lite`, `claude-sonnet-4-20250514`, `gpt-4o`, `gpt-4o-mini`, `llama-4-maverick-17b-128e-instruct`, `claude-3-7-sonnet-20250219` |
| UploadContentType | `image/jpg`, `image/jpeg`, `image/png`, `image/gif`, `image/webp`, `image/svg+xml`, `application/pdf`, `application/msword`, `application/vnd.openxmlformats-officedocument.wordprocessingml.document`, `application/vnd.ms-excel`, `application/vnd.openxmlformats-officedocument.spreadsheetml.sheet`, `text/plain`, `text/csv`, `text/markdown` |
## Response Schemas
diff --git a/skills/open-source/references/quickstart.md b/skills/open-source/references/quickstart.md
index a1af969a2..d7db46cef 100644
--- a/skills/open-source/references/quickstart.md
+++ b/skills/open-source/references/quickstart.md
@@ -64,7 +64,7 @@ import asyncio
load_dotenv()
async def main():
- llm = ChatGoogle(model="gemini-flash-latest")
+ llm = ChatGoogle(model="gemini-3-flash-preview")
agent = Agent(task="Find the number 1 post on Show HN", llm=llm)
await agent.run()
diff --git a/tests/ci/evaluate_tasks.py b/tests/ci/evaluate_tasks.py
index da7709357..5dd14dee3 100644
--- a/tests/ci/evaluate_tasks.py
+++ b/tests/ci/evaluate_tasks.py
@@ -79,7 +79,7 @@ async def run_single_task(task_file):
'explanation': 'Skipped - Google API key not available (fork PR or missing secret)',
}
- judge_llm = ChatGoogle(model='gemini-flash-lite-latest')
+ judge_llm = ChatGoogle(model='gemini-3-flash-preview-lite')
print('[DEBUG] LLMs initialized', file=sys.stderr)
# Each subprocess gets its own profile and session
diff --git a/tests/ci/models/test_llm_google.py b/tests/ci/models/test_llm_google.py
index adb6b779d..96bbb450b 100644
--- a/tests/ci/models/test_llm_google.py
+++ b/tests/ci/models/test_llm_google.py
@@ -4,11 +4,11 @@ from browser_use.llm.google.chat import ChatGoogle
from tests.ci.models.model_test_helper import run_model_button_click_test
-async def test_google_gemini_flash_latest(httpserver):
- """Test Google gemini-flash-latest can click a button."""
+async def test_google_gemini_3_flash_preview(httpserver):
+ """Test Google gemini-3-flash-preview can click a button."""
await run_model_button_click_test(
model_class=ChatGoogle,
- model_name='gemini-flash-latest',
+ model_name='gemini-3-flash-preview',
api_key_env='GOOGLE_API_KEY',
extra_kwargs={},
httpserver=httpserver,