Ollama Python Library

Access models directly at https://ollama.com by configuring a Client with the cloud host and an API key.

1. Create an API key at ollama.com/settings/keys.
2. Set the OLLAMA_API_KEY environment variable.
3. Initialize the Client with host='https://ollama.com' and the appropriate Authorization header.

import os
from ollama import Client

client = Client(
    host='https://ollama.com',
    headers={'Authorization': 'Bearer ' + os.environ.get('OLLAMA_API_KEY')}
)

messages = [
  {
    'role': 'user',
    'content': 'Why is the sky blue?',
  },
]

for part in client.chat('gpt-oss:120b', messages=messages, stream=True):
  print(part.message.content, end='', flush=True)

You can run large cloud models through your local Ollama installation.

1. Sign in locally: ollama signin
2. Pull a cloud model: ollama pull <model_name> (e.g., ollama pull gpt-oss:120b-cloud)
3. Use the Client to interact with it.

from ollama import Client

client = Client()

messages = [
  {
    'role': 'user',
    'content': 'Why is the sky blue?',
  },
]

for part in client.chat('gpt-oss:120b-cloud', messages=messages, stream=True):
  print(part.message.content, end='', flush=True)

Install the library using pip. Ensure you have Ollama installed and running on your system, and that you have pulled the desired model (e.g., ollama pull gemma3) before using the library.

pip install ollama

Errors are raised if requests return an error status or if an error occurs during streaming. Use ollama.ResponseError to catch these exceptions and inspect the error message and status_code.

model = 'does-not-yet-exist'

try:
  ollama.chat(model)
except ollama.ResponseError as e:
  print('Error:', e.error)
  if e.status_code == 404:
    ollama.pull(model)

To receive responses in chunks, set stream=True in the chat function. This returns a generator that you can iterate over.

from ollama import chat

stream = chat(
    model='gemma3',
    messages=[{'role': 'user', 'content': 'Why is the sky blue?'}],
    stream=True,
)

for chunk in stream:
  print(chunk['message']['content'], end='', flush=True)

Use the chat function to send messages to a model. The response can be accessed either as a dictionary or by accessing attributes directly on the ChatResponse object.

from ollama import chat
from ollama import ChatResponse

response: ChatResponse = chat(model='gemma3', messages=[
  {
    'role': 'user',
    'content': 'Why is the sky blue?',
  },
])
print(response['message']['content'])
# or access fields directly from the response object
print(response.message.content)

Instantiate the Client class to specify a custom host or headers. All extra keyword arguments are passed to the underlying httpx.Client.

from ollama import Client
client = Client(
  host='http://localhost:11434',
  headers={'x-some-header': 'some-value'}
)
response = client.chat(model='gemma3', messages=[
  {
    'role': 'user',
    'content': 'Why is the sky blue?',
  },
])

The AsyncClient allows for non-blocking requests. When stream=True is used, the function returns an asynchronous generator.

import asyncio
from ollama import AsyncClient

async def chat():
  message = {'role': 'user', 'content': 'Why is the sky blue?'}
  response = await AsyncClient().chat(model='gemma3', messages=[message])

asyncio.run(chat())

import asyncio
from ollama import AsyncClient

async def chat():
  message = {'role': 'user', 'content': 'Why is the sky blue?'}
  async for part in await AsyncClient().chat(model='gemma3', messages=[message], stream=True):
    print(part['message']['content'], end='', flush=True)

asyncio.run(chat())

The library provides a high-level API mirroring the Ollama REST API. Common methods include:

• chat(model, messages, ...): Chat with a model.
• generate(model, prompt, ...): Generate a response from a prompt.
• list(): List models.
• show(model): Show model information.
• create(model, from_, system, ...): Create a model.
• copy(source, destination): Copy a model.
• delete(model): Delete a model.
• pull(model): Pull a model.
• push(model): Push a model.
• embed(model, input, ...): Generate embeddings (supports single string or list of strings).
• ps(): List running models.