Skip to content
Docs Pydantic Docs

Cerebras

Install

To use CerebrasModel, you need to either install pydantic-ai, or install pydantic-ai-slim with the cerebras optional group:

Terminal
pip install "pydantic-ai-slim[cerebras]"

Configuration

To use Cerebras through their API, go to cloud.cerebras.ai and generate an API key.

For a list of available models, see the Cerebras models documentation.

Environment variable

Once you have the API key, you can set it as an environment variable:

Terminal
export CEREBRAS_API_KEY='your-api-key'

You can then use CerebrasModel by name:

from pydantic_ai import Agent

agent = Agent('cerebras:gpt-oss-120b')
...

Or initialise the model directly with just the model name:

from pydantic_ai import Agent
from pydantic_ai.models.cerebras import CerebrasModel

model = CerebrasModel('gpt-oss-120b')
agent = Agent(model)
...

provider argument

You can provide a custom Provider via the provider argument:

from pydantic_ai import Agent
from pydantic_ai.models.cerebras import CerebrasModel
from pydantic_ai.providers.cerebras import CerebrasProvider

model = CerebrasModel(
    'gpt-oss-120b', provider=CerebrasProvider(api_key='your-api-key')
)
agent = Agent(model)
...

You can also customize the CerebrasProvider with a custom httpx2.AsyncClient:

from httpx2 import AsyncClient

from pydantic_ai import Agent
from pydantic_ai.models.cerebras import CerebrasModel
from pydantic_ai.providers.cerebras import CerebrasProvider

custom_http_client = AsyncClient(timeout=30)
model = CerebrasModel(
    'gpt-oss-120b',
    provider=CerebrasProvider(api_key='your-api-key', http_client=custom_http_client),
)
agent = Agent(model)
...

Thinking

The unified thinking setting follows what Cerebras supports for each model:

  • thinking=False sends reasoning_effort='none' to models that can stop reasoning, such as Qwen 3.8 and GLM-4.7. gpt-oss, Kimi K3 and GLM-5.3 always reason, so it sends their lowest effort, low.
  • Kimi K3 and GLM-5.3 accept only the low, high and max efforts, so the other levels go to the nearest one: minimal to low, medium to high and xhigh to max.
  • On other models, levels other than False aren’t sent, and the model reasons at its default effort.