Available Models & Selection

Duration: 50 min

Bedrock provides access to foundation models from multiple providers. Each model has different strengths, speeds, and costs. This module covers the available models, their capabilities, and how to choose the right one for your use case.

Model Providers & Families

Anthropic Claude

Claude is a family of large language models known for safety, reasoning, and instruction-following.

Claude 3 Family:

• Claude 3 Opus — Most capable, best for complex reasoning (slower, more expensive)
• Claude 3 Sonnet — Balanced speed/capability, recommended for most tasks
• Claude 3 Haiku — Fastest, cheapest, good for simple tasks

import boto3
import json

client = boto3.client('bedrock-runtime', region_name='us-east-1')

# Using Claude 3 Sonnet
response = client.invoke_model(
    modelId='anthropic.claude-3-sonnet-20240229-v1:0',
    body=json.dumps({
        "anthropic_version": "bedrock-2023-06-01",
        "max_tokens": 1024,
        "messages": [
            {"role": "user", "content": "Explain quantum computing"}
        ]
    })
)

print(json.loads(response['body'].read())['content'][0]['text'])

Meta Llama

Open-source models optimized for speed and cost.

Llama 2 & Llama 3:

• Llama 2 70B — Powerful, good for complex tasks
• Llama 3 70B — Improved reasoning and instruction-following
• Llama 3 8B — Lightweight, fast inference

# Using Llama 3 70B
response = client.invoke_model(
    modelId='meta.llama3-70b-instruct-v1:0',
    body=json.dumps({
        "prompt": "What is machine learning?",
        "max_gen_len": 512,
        "temperature": 0.7,
        "top_p": 0.9
    })
)

result = json.loads(response['body'].read())
print(result['generation'])

Mistral AI

Fast, efficient models with strong performance on reasoning tasks.

Mistral Models:

• Mistral Large — Most capable, good for complex tasks
• Mistral 7B — Lightweight, fast
• Mixtral 8x7B — Mixture of experts, efficient

# Using Mistral Large
response = client.invoke_model(
    modelId='mistral.mistral-large-2402-v1:0',
    body=json.dumps({
        "prompt": "Explain RAG systems",
        "max_tokens": 512,
        "temperature": 0.7
    })
)

result = json.loads(response['body'].read())
print(result['outputs'][0]['text'])

Amazon Titan

AWS-native models optimized for Bedrock integration.

Titan Models:

• Titan Text Premier — High-quality text generation
• Titan Text Express — Fast, cost-effective
• Titan Embeddings — Generate vector embeddings for RAG

# Using Titan Text Express
response = client.invoke_model(
    modelId='amazon.titan-text-express-v1:0',
    body=json.dumps({
        "inputText": "Summarize AWS Bedrock",
        "textGenerationConfig": {
            "maxTokenCount": 512,
            "temperature": 0.7,
            "topP": 0.9
        }
    })
)

result = json.loads(response['body'].read())
print(result['results'][0]['outputText'])

Stability AI

Image generation and manipulation models.

Stable Diffusion:

• Stable Diffusion 3 — Latest, highest quality
• Stable Diffusion XL — Fast, good quality
• Stable Diffusion 1.6 — Lightweight

import base64

# Generate an image
response = client.invoke_model(
    modelId='stability.stable-diffusion-xl-v1:0',
    body=json.dumps({
        "text_prompts": [
            {"text": "A futuristic AI assistant", "weight": 1.0}
        ],
        "cfg_scale": 10,
        "steps": 50,
        "seed": 0
    })
)

result = json.loads(response['body'].read())
image_data = base64.b64decode(result['artifacts'][0]['base64'])
with open('generated_image.png', 'wb') as f:
    f.write(image_data)

Model Comparison Matrix

How to Choose a Model

For customer support chatbots:

• Use Claude 3 Sonnet or Llama 3 70B
• Fast response time, good instruction-following
• Cost-effective at scale

For complex reasoning & analysis:

• Use Claude 3 Opus or Mistral Large
• Best for multi-step problem solving
• Accept slower response times

For code generation:

• Use Claude 3 Sonnet or Llama 3 70B
• Both excel at code tasks
• Sonnet slightly better for complex logic

For high-volume, cost-sensitive tasks:

• Use Llama 3 8B or Mistral 7B
• Fastest inference, lowest cost
• Good for simple classification, summarization

For image generation:

• Use Stable Diffusion 3 for quality
• Use Stable Diffusion XL for speed/cost balance

Listing Available Models

# List all available models
aws bedrock list-foundation-models --region us-east-1

# Filter by provider
aws bedrock list-foundation-models \
  --by-provider anthropic \
  --region us-east-1

# Filter by capability
aws bedrock list-foundation-models \
  --by-customization-type FINE_TUNING \
  --region us-east-1

Model IDs Reference

{
  "claude": {
    "opus": "anthropic.claude-3-opus-20240229-v1:0",
    "sonnet": "anthropic.claude-3-sonnet-20240229-v1:0",
    "haiku": "anthropic.claude-3-haiku-20240307-v1:0"
  },
  "llama": {
    "llama3_70b": "meta.llama3-70b-instruct-v1:0",
    "llama3_8b": "meta.llama3-8b-instruct-v1:0",
    "llama2_70b": "meta.llama2-70b-chat-v1:0"
  },
  "mistral": {
    "large": "mistral.mistral-large-2402-v1:0",
    "7b": "mistral.mistral-7b-instruct-v0:2"
  },
  "titan": {
    "text_premier": "amazon.titan-text-premier-v1:0",
    "text_express": "amazon.titan-text-express-v1:0",
    "embeddings": "amazon.titan-embed-text-v2:0"
  }
}