Nova Reel (Tokyo)

Foundation model available in Tokyo. Direct inference without cross-region routing.

Provider: All Amazon models | AWS Bedrock

Inference regions: ap-northeast-1

API Endpoint

https://bedrock-runtime.ap-northeast-1.amazonaws.com

Quick Start (Python)

Install: pip install boto3

import boto3

client = boto3.client("bedrock-runtime", region_name="ap-northeast-1")

response = client.converse(
    modelId="amazon.nova-reel-v1",
    messages=[
        {
            "role": "user",
            "content": [{"text": "Hello, how are you?"}],
        }
    ],
    inferenceConfig={
        "maxTokens": 1024,
        "temperature": 0.7,
    },
)

print(response["output"]["message"]["content"][0]["text"])

Additional examples: Basic invoke, Streaming

Supported Parameters

Parameter	Type	Description
taskType	enum	Type of video generation task.
durationSeconds	integer	Duration of the generated video in seconds. (6–24) Default: 6.
fps	integer	Frames per second. Default: 24.
dimension	enum	Output video dimensions.

Feature Guides

On-Demand Inference

Default mode. Pay per token with no upfront commitment.

Documentation

Async Video Generation

Nova Reel uses asynchronous inference. Submit a job and poll for results.

Call StartAsyncInvoke with the video generation parameters
Receive an invocationArn in the response
Poll GetAsyncInvoke with the ARN until status is COMPLETED
Retrieve the video from the output S3 location

Documentation