AWS Bedrock
Our foundation models models are available through Amazon Bedrock. AI21 Models on Bedrock are hosted on AWS and usage is per token. You do not need to provision or manage these models yourself.
Limited region support
Bedrock is currently supported only in the region us-east-1.
Usage
You'll be using the AI21 Python SDK to access all AI21 models.
- Create an Amazon Bedrock instance and enable access to AI21 models on Bedrock.
- Install or update the AI21 Studio Python SDK with the latest AWS integration:
pip install -U "ai21[AWS] >= 2.13.0" - Create your
AI21BedrockClientclient.- You can either use your current environment's AWS session keys, or else create and pass in a
boto3session explicitly, if you want to manage your keys explicitly. You do not need to use AI21 client keys, only AWS keys, because AWS handles your AI21 authorization for you. - The only region currently supported is
us-east-1.
- You can either use your current environment's AWS session keys, or else create and pass in a
- Follow the same usage as described in the AI21 Studio Python SDK Guide, but be aware that only certain models are available on Bedrock.
Examples
from ai21 import AI21BedrockClient, BedrockModelID
from ai21.models.chat import ChatMessage
client = AI21BedrockClient(region='us-east-1') # region is optional, as you can use the env variable instead
messages = [
ChatMessage(content="You are a helpful assistant", role="system"),
ChatMessage(content="What is the meaning of life?", role="user")
]
response = client.chat.completions.create(
messages=messages,
model_id=BedrockModelID.JAMBA_INSTRUCT_V1,
)
from ai21 import AI21BedrockClient, BedrockModelID
from ai21.models.chat import ChatMessage
boto_session = boto3.Session(
region_name="us-east-1",
aws_access_key_id=aws_access_key_id,
aws_secret_access_key=aws_secret_access_key,
aws_session_tokeΧ=aws_session_token,
)
client = AI21BedrockClient(session=boto_session, model_id=BedrockModelID.JAMBA_INSTRUCT_V1)
messages = [
ChatMessage(content="You are a helpful assistant", role="system"),
ChatMessage(content="What is the meaning of life?", role="user")
]
response = client.chat.completions.create(
messages=messages,
model_id=BedrockModelID.JAMBA_INSTRUCT_V1,
)
More information
- See the list of available models on Bedrock .
- Learn how to use the SDK
- Learn more about Amazon Bedrock.
Updated 4 months ago