OpenAI | Agno

GPT models are used as the default LLM by Agents. OpenAI supports a variety of models. See the full list.

We recommend experimenting to find the best-suited model for your use-case. Here are some general recommendations:

gpt-5.4 is good for most general use-cases.
gpt-5.4-mini is good for smaller tasks and faster inference.
o4-mini and o3 are good for complex reasoning and multi-step tasks.

OpenAI has tier-based rate limits. See the rate limit docs.

Installation

uv pip install -U agno openai

Authentication

Set your OPENAI_API_KEY environment variable. You can get one from OpenAI here.

export OPENAI_API_KEY="YOUR_API_KEY"

Example

Use OpenAIChat with your Agent:

from agno.agent import Agent
from agno.models.openai import OpenAIChat

agent = Agent(
    model=OpenAIChat(id="gpt-5.4-mini"),
    markdown=True
)

# Print the response in the terminal
agent.print_response("Share a 2 sentence horror story.")

Prompt caching

OpenAI applies prompt caching automatically to eligible prompt prefixes. This is separate from Agno's local response cache. Read more about how OpenAI handles caching in their docs.

View more examples here.

Parameters

See the OpenAI Chat Completions reference. The table lists Agno defaults: None generally leaves a setting to the API. Not every model accepts every parameter or value.

Parameter	Type	Default	Description
`id`	`str`	`"gpt-5.4-mini"`	The id of the OpenAI model to use
`name`	`str`	`"OpenAIChat"`	The name of the model
`provider`	`str`	`"OpenAI"`	The provider of the model
`collect_metrics_on_completion`	`bool`	`False`	Collect token metrics only from the final streaming chunk (for providers with cumulative token counts)
`store`	`Optional[bool]`	`None`	Store completion output for use in distillation or evals
`reasoning_effort`	`Optional[str]`	`None`	Reasoning effort; accepted values depend on the model (for example `none`, `minimal`, `low`, `medium`, `high`, `xhigh`, `max`)
`verbosity`	`Optional[Literal["low", "medium", "high"]]`	`None`	Controls verbosity level of reasoning models
`metadata`	`Optional[Dict[str, Any]]`	`None`	Developer-defined metadata to associate with the completion
`frequency_penalty`	`Optional[float]`	`None`	Penalizes new tokens based on their frequency in the text so far (-2.0 to 2.0)
`logit_bias`	`Optional[Any]`	`None`	Modifies the likelihood of specified tokens appearing in the completion
`logprobs`	`Optional[bool]`	`None`	Whether to return log probabilities of the output tokens
`top_logprobs`	`Optional[int]`	`None`	Number of most likely tokens to return log probabilities for (0 to 20)
`max_tokens`	`Optional[int]`	`None`	Maximum number of tokens to generate (deprecated, use max_completion_tokens)
`max_completion_tokens`	`Optional[int]`	`None`	Maximum number of completion tokens to generate
`modalities`	`Optional[List[str]]`	`None`	List of modalities to use ("text" and/or "audio")
`audio`	`Optional[Dict[str, Any]]`	`None`	Audio configuration (e.g., `{"voice": "alloy", "format": "wav"}`)
`presence_penalty`	`Optional[float]`	`None`	Penalizes new tokens based on whether they appear in the text so far (-2.0 to 2.0)
`seed`	`Optional[int]`	`None`	Best-effort reproducibility where supported; determinism is not guaranteed
`stop`	`Optional[Union[str, List[str]]]`	`None`	Up to 4 sequences where the API will stop generating further tokens
`temperature`	`Optional[float]`	`None`	Controls randomness in the model's output (0.0 to 2.0)
`user`	`Optional[str]`	`None`	A unique identifier representing your end-user
`top_p`	`Optional[float]`	`None`	Controls diversity via nucleus sampling (0.0 to 1.0)
`service_tier`	`Optional[str]`	`None`	Service tier such as `auto`, `default`, `flex`, `priority`, `scale`, or `fast`; availability depends on model and endpoint
`strict_output`	`bool`	`True`	Controls schema adherence for structured outputs
`extra_headers`	`Optional[Any]`	`None`	Additional headers to include in requests
`extra_query`	`Optional[Any]`	`None`	Additional query parameters to include in requests
`extra_body`	`Optional[Any]`	`None`	Additional body parameters to include in requests
`request_params`	`Optional[Dict[str, Any]]`	`None`	Additional parameters to include in the request
`role_map`	`Optional[Dict[str, str]]`	`None`	Mapping of message roles to OpenAI roles
`api_key`	`Optional[str]`	`None`	The API key for authenticating with OpenAI (defaults to OPENAI_API_KEY env var)
`organization`	`Optional[str]`	`None`	The organization ID to use for requests
`base_url`	`Optional[Union[str, httpx.URL]]`	`None`	The base URL for the OpenAI API
`timeout`	`Optional[float]`	`None`	Request timeout in seconds
`max_retries`	`Optional[int]`	`None`	Maximum number of retries for failed requests
`default_headers`	`Optional[Any]`	`None`	Default headers to include in all requests
`default_query`	`Optional[Any]`	`None`	Default query parameters to include in all requests
`http_client`	`Optional[Union[httpx.Client, httpx.AsyncClient]]`	`None`	HTTP client instance for making requests
`client_params`	`Optional[Dict[str, Any]]`	`None`	Additional parameters for client configuration

OpenAIChat is a subclass of the Model class and has access to the same params.