​
Overview

OpenOperator supports various model providers. Here’s how to configure and use the most popular ones.

​
Model Recommendations

We have yet to test performance across all models. Obviously the more capabable the model the better the performance. Thus it’s recommended to use any of openai/gpt-4o, deepseek/deepseek-chat, deepseek/deepseek-reasoner, gemini/gemini-2.0-flash-thinking-exp-01-21, gemini/gemini-2.0-flash-exp or anthopic/claude-3-5-sonnet-20241022. We also support local models via ollama, like ollama/qwen2.5, ollama/deepseek-r1 or ollama/llama3.3, but be aware that small models often return the wrong output structure, which leads to parsing errors. We believe that local models will improve significantly this year.
All models require their respective API keys. Make sure to set them in your environment variables before running the agent.

​
Supported Models

Here’s a non exhaustive list of supported model providers. For the full list of supported model providers see https://github.com/j0yk1ll/openoperator/blob/main/openoperator/llm.py

​
OpenAI

OpenAI’s GPT-4o models are recommended for best performance. 
from openoperator import Agent, LLM

# Initialize the model
llm = LLM(
    model="openai/gpt-4o",
    temperature=0.0,
)

# Create agent with the model
agent = Agent(llm=llm)
Required environment variables:
.env
OPENAI_API_KEY=

​
Anthropic

from openoperator import Agent, LLM

# Initialize the model
llm = LLM(
    model_name="anthopic/claude-3-5-sonnet-20241022",
    temperature=0.0,
    timeout=100, # Increase for complex tasks
)

# Create agent with the model
agent = Agent(llm=llm)
And add the variable:
.env
ANTHROPIC_API_KEY=

​
Azure OpenAI

from openoperator import Agent, LLM

# Initialize the model
llm = LLM(
    model="azure/gpt-4o",
    api_version='2024-10-21'
)

# Create agent with the model
agent = Agent(llm=llm)
Required environment variables:
.env
AZURE_OPENAI_ENDPOINT=https://your-endpoint.openai.azure.com/
AZURE_OPENAI_API_KEY=

​
Groq

from openoperator import Agent, LLM

# Initialize the model
llm = LLM(
    model="groq/llama-3.3-70b-versatile"
)

# Create agent with the model
agent = Agent(llm=llm)
Required environment variables:
.env
GROQ_API_KEY=

​
Gemini

from openoperator import Agent, LLM

# Initialize the model
llm = LLM(model='gemini/gemini-2.0-flash-exp')

# Create agent with the model
agent = Agent(llm=llm)
Required environment variables:
.env
GEMINI_API_KEY=

​
DeepSeek-V3

The community likes DeepSeek-V3 for its low price, no rate limits, open-source nature, and good performance. The example is available here. 
from openoperator import Agent, LLM

# Initialize the model
llm=LLM(model='deepseek/deepseek-chat')

# Create agent with the model
agent = Agent(
    llm=llm,
    use_vision=False
)
Required environment variables:
.env
DEEPSEEK_API_KEY=

​
DeepSeek-R1

We support DeepSeek-R1. Its not fully tested yet, more and more functionality will be added, like e.g. the output of it’s reasoning content. The example is available here. It does not support vision. The model is open-source so you could also use it with Ollama, but we have not tested it. 
from openoperator import Agent, LLM


# Initialize the model
llm=LLM(model='deepseek/deepseek-reasoner')

# Create agent with the model
agent = Agent(
    llm=llm,
    use_vision=False
)
Required environment variables:
.env
DEEPSEEK_API_KEY=

​
Ollama

You can use Ollama to easily run local models.
  1. Download Ollama from here
  2. Run ollama pull model_name. Preferably pick a model which supports tool-calling from here
  3. Run ollama start
from openoperator import Agent, LLM

# Initialize the model
llm=LLM(model="ollama/qwen2.5", num_ctx=32000)

# Create agent with the model
agent = Agent(
    llm=llm
)
Required environment variables: None!