Last refreshed 2026-07-09. Next refresh: weekly.
Why use Llama 3.1 405B Instruct on AWS Bedrock?
AWS Bedrock offers Llama 3.1 405B Instruct with pay-as-you-go pricing at $2.40/1M input tokens. AWS Bedrock is Amazon's fully managed foundation-model service, providing unified API access to top models from Anthropic, Meta, Mistral, and other leading AI labs with built-in tools for RAG, fine-tuning, and AI agent development.
Compare Llama 3.1 405B Instruct across 11 providers to find the best fit for your use caseSetup recipe
Python + curlpip install boto3export AWS_ACCESS_KEY_ID=...import boto3
client = boto3.client("bedrock-runtime", region_name="us-east-1")
response = client.converse(
modelId="llama3.1-405b-instruct",llama3.1-405b-instructRequest example
import boto3
# Reads AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION from env
client = boto3.client("bedrock-runtime", region_name="us-east-1")
response = client.converse(
modelId="llama3.1-405b-instruct",
messages=[{
"role": "user",
"content": [{"text": "Hello"}]
}]
)
print(response["output"]["message"]["content"][0]["text"])Gotchas
- Use Amazon Bedrock model IDs, e.g. "anthropic.claude-3-opus-20240229-v1:0" for on-demand, or cross-region inference profile IDs like "us.anthropic.claude-opus-4-7-20251101-v1:0". These differ from the public model slug.
- The endpoint template includes a region segment; set the same region in your SDK/client configuration.
- The examples expect AWS_ACCESS_KEY_ID; rename it only if your application config maps the new variable.
Compare Llama 3.1 405B Instruct Across Providers
| Provider | Input (per 1M) | Output (per 1M) |
|---|---|---|
| OctoAI API (Deprecated) | — | — |
| Together AI | $5.00 | $15.00 |
| Fireworks AI | $3.00 | $3.00 |
| IBM watsonx | $3.00 | $9.00 |
| Scale AI GenAI Platform | — | — |
Pricing
| Type | Price (per 1M) |
|---|---|
| Input tokens | $2.40 |
| Output tokens | $2.40 |
Capabilities
About Llama 3.1 405B Instruct
Llama 3.1 405B Instruct is Meta's advanced large language model released on July 23, 2024, featuring 405 billion parameters. It utilizes an optimized transformer architecture with supervised fine-tuning and reinforcement learning for enhanced instruction-following capabilities. The model supports multiple languages, was trained on 15 trillion tokens, and fine-tuned with 25 million synthetic examples. It excels in multilingual dialogue and text generation, making it ideal for assistant-like applications. Llama 3.1 incorporates robust safety measures and ethical considerations, outperforming many existing models on various industry benchmarks.