Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.
Python ai-agentsai-toolsbrowser-agentbrowser-automationbrowser-usecdpcloud-browsercomputer-useheadless-browserllmmcpplaywrightpythonremote-browserstealth-browserweb-automationweb-scraping
refactor(cli): use SDK verified model lists instead of hardcoded prefixes
ShawnPanaaf9383a2c7f2
diff --git a/browser_use/llm/browser_use/chat.py b/browser_use/llm/browser_use/chat.pyindex 26b73b12d..ead81ba24 100644--- a/browser_use/llm/browser_use/chat.py+++ b/browser_use/llm/browser_use/chat.py@@ -27,6 +27,9 @@ logger = logging.getLogger(__name__)# HTTP status codes that should trigger a retryRETRYABLE_STATUS_CODES = {429, 500, 502, 503, 504}+# Valid model names for ChatBrowserUse+VALID_BROWSER_USE_MODELS = {'bu-latest', 'bu-1-0', 'bu-2-0'}+class ChatBrowserUse(BaseChatModel):"""@@ -60,7 +63,6 @@ class ChatBrowserUse(BaseChatModel):model: Model name to use. Options:- 'bu-latest' or 'bu-1-0': Default model- 'bu-2-0': Latest premium model- - 'browser-use/bu-30b-a3b-preview': Browser Use Open Source Modelapi_key: API key for browser-use cloud. Defaults to BROWSER_USE_API_KEY env var.base_url: Base URL for the API. Defaults to BROWSER_USE_LLM_URL env var or production URL.timeout: Request timeout in seconds.@@ -68,11 +70,9 @@ class ChatBrowserUse(BaseChatModel):retry_base_delay: Base delay in seconds for exponential backoff (default: 1.0).retry_max_delay: Maximum delay in seconds between retries (default: 60.0)."""- # Validate model name - allow bu-* and browser-use/* patterns- valid_models = ['bu-latest', 'bu-1-0', 'bu-2-0']- is_valid = model in valid_models or model.startswith('browser-use/')- if not is_valid:- raise ValueError(f"Invalid model: '{model}'. Must be one of {valid_models} or start with 'browser-use/'")+ # Validate model name+ if model not in VALID_BROWSER_USE_MODELS:+ raise ValueError(f"Invalid model: '{model}'. Must be one of {VALID_BROWSER_USE_MODELS}")# Normalize bu-latest to bu-1-0 for default modelsif model == 'bu-latest':diff --git a/browser_use/skill_cli/commands/agent.py b/browser_use/skill_cli/commands/agent.pyindex e89e345f0..81ed94af1 100644--- a/browser_use/skill_cli/commands/agent.py+++ b/browser_use/skill_cli/commands/agent.py@@ -244,31 +244,54 @@ async def _handle_local_task(session: SessionInfo, params: dict[str, Any]) -> An}-def _get_provider_from_model(model: str) -> str | None:- """Determine the provider from a model name.+def _get_verified_models() -> dict[str, set[str]]:+ """Extract verified model names from SDK sources of truth."""+ import typing- Returns 'openai', 'anthropic', 'google', or None if unknown.- """- model_lower = model.lower()+ from anthropic.types.model_param import ModelParam+ from openai.types.shared.chat_model import ChatModel++ from browser_use.llm.browser_use.chat import VALID_BROWSER_USE_MODELS+ from browser_use.llm.google.chat import VerifiedGeminiModels++ # OpenAI: ChatModel is a Literal type+ openai_models = set(typing.get_args(ChatModel))++ # Anthropic: ModelParam is Union[Literal[...], str] - extract the Literal+ anthropic_literal = typing.get_args(ModelParam)[0]+ anthropic_models = set(typing.get_args(anthropic_literal))++ # Google: VerifiedGeminiModels Literal+ google_models = set(typing.get_args(VerifiedGeminiModels))++ # Browser-Use: VALID_BROWSER_USE_MODELS constant+ browser_use_models = VALID_BROWSER_USE_MODELS++ return {+ 'openai': openai_models,+ 'anthropic': anthropic_models,+ 'google': google_models,+ 'browser-use': browser_use_models,+ }- # Check for explicit provider prefix (e.g., "openai/gpt-4o")- if '/' in model:- provider = model.split('/')[0].lower()- if provider in ('openai', 'anthropic', 'google'):- return provider- # Infer from model name patterns- if model_lower.startswith('gpt-') or model_lower.startswith('o1') or model_lower.startswith('o3'):- return 'openai'- elif model_lower.startswith('claude-'):- return 'anthropic'- elif model_lower.startswith('gemini-'):- return 'google'+_VERIFIED_MODELS: dict[str, set[str]] | None = None+++def _get_provider_for_model(model: str) -> str | None:+ """Determine the provider by checking SDK verified model lists."""+ global _VERIFIED_MODELS+ if _VERIFIED_MODELS is None:+ _VERIFIED_MODELS = _get_verified_models()++ for provider, models in _VERIFIED_MODELS.items():+ if model in models:+ return providerreturn None-async def get_llm(model: str | None = None) -> Any:+def get_llm(model: str | None = None) -> Any:"""Get LLM instance from environment configuration.Args:@@ -276,103 +299,37 @@ async def get_llm(model: str | None = None) -> Any:the appropriate provider for that model. If not provided,auto-detects from available API keys.- Supported model patterns:- - gpt-*, o1*, o3* → OpenAI (requires OPENAI_API_KEY)- - claude-* → Anthropic (requires ANTHROPIC_API_KEY)- - gemini-* → Google (requires GOOGLE_API_KEY)- - openai/model, anthropic/model, google/model → Explicit provider+ Supported providers: OpenAI, Anthropic, Google, Browser-Use.+ Model names are validated against each SDK's verified model list."""- # If model specified, use that provider- if model:- provider = _get_provider_from_model(model)+ from browser_use.llm import ChatAnthropic, ChatBrowserUse, ChatGoogle, ChatOpenAI- # Strip provider prefix if present (e.g., "openai/gpt-4o" → "gpt-4o")- model_name = model.split('/')[-1] if '/' in model else model+ if model:+ provider = _get_provider_for_model(model)if provider == 'openai':- if not os.environ.get('OPENAI_API_KEY'):- logger.warning(f'Model {model} requires OPENAI_API_KEY')- return None- try:- from langchain_openai import ChatOpenAI # type: ignore[import-not-found]-- return ChatOpenAI(model=model_name) # type: ignore[return-value]- except ImportError:- logger.warning('langchain-openai not installed')- return None-+ return ChatOpenAI(model=model)elif provider == 'anthropic':- if not os.environ.get('ANTHROPIC_API_KEY'):- logger.warning(f'Model {model} requires ANTHROPIC_API_KEY')- return None- try:- from langchain_anthropic import ChatAnthropic # type: ignore[import-not-found]-- return ChatAnthropic(model=model_name) # type: ignore[return-value]- except ImportError:- logger.warning('langchain-anthropic not installed')- return None-+ return ChatAnthropic(model=model)elif provider == 'google':- if not os.environ.get('GOOGLE_API_KEY'):- logger.warning(f'Model {model} requires GOOGLE_API_KEY')- return None- try:- from langchain_google_genai import ChatGoogleGenerativeAI # type: ignore[import-not-found]-- return ChatGoogleGenerativeAI(model=model_name) # type: ignore[return-value]- except ImportError:- logger.warning('langchain-google-genai not installed')- return None-+ return ChatGoogle(model=model)+ elif provider == 'browser-use':+ return ChatBrowserUse(model=model)else:- logger.warning(f'Unknown model provider for: {model}')+ logger.warning(f'Unknown model: {model}. Not in any verified model list.')return None# No model specified - auto-detect from available API keys-- # Try ChatBrowserUse first (optimized for browser automation)if os.environ.get('BROWSER_USE_API_KEY'):- try:- from browser_use.llm import ChatBrowserUse+ return ChatBrowserUse()- return ChatBrowserUse() # type: ignore[return-value]- except ImportError:- pass-- # Try OpenAIif os.environ.get('OPENAI_API_KEY'):- try:- from langchain_openai import ChatOpenAI # type: ignore[import-not-found]+ return ChatOpenAI(model='o3')- return ChatOpenAI(model='gpt-4o') # type: ignore[return-value]- except ImportError:- pass-- # Try Anthropicif os.environ.get('ANTHROPIC_API_KEY'):- try:- from langchain_anthropic import ChatAnthropic # type: ignore[import-not-found]+ return ChatAnthropic(model='claude-sonnet-4-0')- return ChatAnthropic(model='claude-sonnet-4-20250514') # type: ignore[return-value]- except ImportError:- pass-- # Try Googleif os.environ.get('GOOGLE_API_KEY'):- try:- from langchain_google_genai import ChatGoogleGenerativeAI # type: ignore[import-not-found]-- return ChatGoogleGenerativeAI(model='gemini-2.0-flash') # type: ignore[return-value]- except ImportError:- pass-- # Try to use browser-use's default LLM setup- try:- from browser_use.llm import get_default_llm-- return get_default_llm() # type: ignore[return-value]- except Exception:- pass+ return ChatGoogle(model='gemini-flash-latest')return None