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
Simplify models
Magnus Müller76d8d99ab62a
diff --git a/docs/customize/supported-models.mdx b/docs/customize/supported-models.mdxindex 997494dd3..a5f8bcd39 100644--- a/docs/customize/supported-models.mdx+++ b/docs/customize/supported-models.mdx@@ -1,24 +1,23 @@---title: "Supported Models"-description: "Using different chat providers with Browser Use"+description: "Choose your favorite LLM"icon: "robot"----## Model Recommendations+### Recommendations-We recommend using `O3` for the best performance. The best price to performance can be achieved using `gemini-2.0-flash-exp`.+- Best accuracy: `O3`+- Fastest: `llama4` on groq+- Balanced: fast + cheap + clever: `gemini-2.5-flash`-## Supported Models--In addition to all the models below, we support all other models that can be called via OpenAI compatible API (deepseek, novita, x, qwen). We are open to PRs for more providers.+Find full examples built for you in github [examples/models](https://github.com/browser-use/browser-use/tree/main/examples/models).### OpenAI`O3` model is recommended for best performance.```python-from browser_use.llm import ChatOpenAI-from browser_use import Agent+from browser_use import Agent, ChatOpenAI# Initialize the modelllm = ChatOpenAI(@@ -47,8 +46,7 @@ OPENAI_API_KEY=### Anthropic```python-from browser_use.llm import ChatAnthropic-from browser_use import Agent+from browser_use import Agent, ChatAnthropic# Initialize the modelllm = ChatAnthropic(@@ -71,8 +69,7 @@ ANTHROPIC_API_KEY=### Azure OpenAI```python-from browser_use.llm import ChatAzureOpenAI-from browser_use import Agent+from browser_use import Agent, ChatAzureOpenAIfrom pydantic import SecretStrimport os@@ -100,8 +97,7 @@ AZURE_OPENAI_API_KEY=> [!IMPORTANT] `GEMINI_API_KEY` was the old environment var name, it should be called `GOOGLE_API_KEY` as of 2025-05.```python-from browser_use.llm import ChatGoogle-from browser_use import Agent+from browser_use import Agent, ChatGooglefrom dotenv import load_dotenv# Read GOOGLE_API_KEY into env@@ -130,8 +126,7 @@ AWS Bedrock provides access to multiple model providers through a single API. We#### General AWS Bedrock (supports all providers)```python-from browser_use.llm import ChatAWSBedrock-from browser_use import Agent+from browser_use import Agent, ChatAWSBedrock# Works with any Bedrock model (Anthropic, Meta, AI21, etc.)llm = ChatAWSBedrock(@@ -149,8 +144,7 @@ agent = Agent(#### Anthropic Claude via AWS Bedrock (convenience class)```python-from browser_use.llm import ChatAnthropicBedrock-from browser_use import Agent+from browser_use import Agent, ChatAnthropicBedrock# Anthropic-specific class with Claude defaultsllm = ChatAnthropicBedrock(@@ -186,8 +180,7 @@ You can also use AWS profiles or IAM roles instead of environment variables. The## Groq```python-from browser_use.llm import ChatGroq-from browser_use import Agent+from browser_use import Agent, ChatGroqllm = ChatGroq(model="meta-llama/llama-4-maverick-17b-128e-instruct")@@ -206,16 +199,15 @@ GROQ_API_KEY=## Ollama```python-from browser_use.llm import ChatOllama-from browser_use import Agent+from browser_use import Agent, ChatOllamallm = ChatOllama(model="llama3.1:8b")```-## Migration Guides+## Langchain-### From Langchain+[Example](https://github.com/browser-use/browser-use/blob/main/examples/models/langchain) on how to use Langchain with Browser Use.-To migrate the Langchain based code, just replace `from langchain_openai import ChatOpenAI` with `from browser_use.llm import ChatOpenAI` etc. The methods should be compatible(ish).+## Other models (DeepSeek, Novita, X, Qwen...)-We also made and example [here](https://github.com/browser-use/browser-use/blob/main/examples/models/langchain) to help you stay with Langchain in case your workflow requires it.+We support all other models that can be called via OpenAI compatible API. We are open to PRs for more providers.