Amazon Bedrock Verified
What is Amazon Bedrock?
Amazon Bedrock is a fully managed AWS service that provides API access to foundation models from Anthropic, Meta, Mistral, Amazon, and partners - without managing infrastructure. Teams use it to build RAG pipelines, enterprise copilots, and multimodal applications while accessing Bedrock privately from their AWS VPC through AWS PrivateLink interface endpoints. It is a primary route for enterprises that need IAM-scoped access, CloudTrail audit logs, and regional data-processing controls alongside foundation-model capabilities.
Pricing
On-demand pricing, USD per 1K tokens, as of April 2026 - verify current rates at aws.amazon.com/bedrock/pricing.
| Model | Input | Output |
|---|---|---|
| Amazon Nova Micro | $0.000035 | $0.000140 |
| Amazon Nova Lite | $0.000060 | $0.000240 |
| Amazon Nova Pro | $0.000800 | $0.003200 |
| Claude Haiku 3.5 (via Bedrock) | $0.000800 | $0.004000 |
| Claude Sonnet 4 (via Bedrock) | $0.003000 | $0.015000 |
| Claude Opus 4 (via Bedrock) | $0.015000 | $0.075000 |
| Llama 4 Scout (via Bedrock) | $0.000170 | $0.000170 |
| Mistral Large 2 (via Bedrock) | $0.002000 | $0.006000 |
Provisioned throughput is available for committed workloads - price varies by model unit and term length. Cross-region inference may add latency; verify model availability per region before architecture sign-off.
Key insights
Concrete technical or product signals.
- Teams already on AWS often standardize on Bedrock for procurement, private networking, and CloudTrail-style auditing.
- Model availability and default versions vary by region—verify the model ID list for your account before architecture sign-off.
Use cases
Where this shines in production.
- Enterprise copilots with IAM-scoped access to approved models
- RAG pipelines that must stay inside a VPC with KMS-encrypted data
- Multimodal apps using Nova-class models without self-hosting GPUs
Limitations & trade-offs
What to watch for.
- Cold-start latency and quota tiers differ by model; load-test your expected concurrency.
- Cross-region failover and model deprecation notices require operational runbooks.
Models referenced
Declared model dependencies or integrations.
Claude Opus 4, Claude Sonnet 4, Claude Haiku 3.5, Claude 3.5 Sonnet, Amazon Nova Pro, Amazon Nova Lite, Amazon Nova Micro, Amazon Titan Text Premier, Llama 4 Scout, Llama 4 Maverick, Llama 3.1 405B Instruct, Mistral Large 2, Stable Diffusion XL
Availability varies by model and AWS Region. Verify the Bedrock model catalog before deployment.
Related prompts
Hand-picked or latest prompt templates.
Prompt
RAG Pipeline System Prompt Template
A production system-prompt template for retrieval-grounded answers with citation, access-control, and empty-retrieval handling rules.
Prompt
API Error Triage Workflow
A structured approach to identifying, categorizing, and resolving API errors in production systems.
Prompt
Model Evaluation Rubric for Production LLMs
A repeatable rubric for comparing production LLM candidates across quality, latency, cost, tool use, safety, and operational fit.
Prompt
Bedrock Converse API Integration Pattern
An implementation checklist for Bedrock Converse API integrations covering model IDs, retries, streaming, tool calls, IAM, and observability.
Prompt
Vector Embedding Pipeline for Enterprise RAG
A design template for enterprise embedding pipelines covering chunking, metadata, tenancy, indexing, refreshes, and retrieval evaluation.
Looking for a tighter match? Search the prompt library.
Bedrock vs Azure OpenAI - quick decision guide
| Dimension | Amazon Bedrock | Azure OpenAI |
|---|---|---|
| Model catalog | Anthropic, Meta, Mistral, Amazon, AI21 | OpenAI model family, including GPT-4o, o3, and DALL-E 3 |
| Cloud requirement | AWS-native integration | Azure-native integration |
| Identity & access | IAM roles, SCPs, VPC endpoints | Microsoft Entra ID, Azure RBAC, private endpoints |
| Best fit | AWS-native teams needing multi-provider model access | Microsoft and Azure shops needing OpenAI models |
| Pricing model | Per-token on-demand or provisioned throughput | Per-token on-demand or PTU provisioned throughput |
If your team is on AWS and needs Claude or Llama, Bedrock is the path of least resistance. If you need GPT-4o or o3 specifically inside Azure, Azure OpenAI is the managed Azure route.
Related
Comparisons, platforms, and models teams often view next.