Amazon Bedrock, SageMaker AI & EC2 GPU Instances
Hyperscaler cloud
Provided by Amazon Web Services, Inc.
Offering overview
Three layers define AI hosting on AWS. Amazon Bedrock is a serverless API over foundation models from Anthropic, Amazon Nova, Meta Llama, Mistral, DeepSeek, Qwen, OpenAI gpt-oss, Cohere, AI21, Google Gemma, xAI Grok, Writer and others, with custom model import, knowledge bases for RAG and agent tooling. SageMaker AI covers notebooks, managed training, fine-tuning and real-time or batch inference endpoints. Beneath both sit EC2 accelerated instances built on NVIDIA GPUs (G4dn, G5, P4, P5 and newer) and Amazon's own Trainium and Inferentia chips. Strengths include the breadth of regions, deep integration with IAM, VPC and storage, provisioned throughput options, and on-premises extensions through Outposts and Local Zones. Prices differ by region and model, and batch inference is discounted. This is the default choice for enterprises already running on AWS and for teams that want one vendor for data, training and serving. Entry pricing on Bedrock begins around USD 0.04 per 1M input tokens for Google Gemma 3 4B, and SageMaker ml.g4dn.xlarge instances are listed from roughly USD 0.74 per hour.
- Management
- Not stated
- Infrastructure
- Not stated
- Pricing model
- Not stated
- Support
- Not stated
- SLA
- Not stated
- Best for
- Model Inference, Model Training, Fine-Tuning, AI Agents, AI Applications, Vector Search, Batch Processing, Edge Inference, Research And Experimentation
Technical specifications
- Containers
- Unknown
- Kubernetes
- Unknown
- Root access
- Unknown
- API access
- Unknown
- Autoscaling
- Unknown
- Persistent processes
- Unknown
AI capabilities
- Model serving
- Anthropic Claude, Amazon Nova, Llama, Mistral, DeepSeek, Qwen, OpenAI gpt-oss, Cohere Command, AI21 Jamba, Google Gemma, xAI Grok, NVIDIA Nemotron, MiniMax, Kimi, GLM, Stability AI, Writer Palmyra, Luma AI, TwelveLabs
- AI frameworks
- Not stated
- Inference endpoints
- Unknown
- Fine-tuning
- Unknown
- Vector database
- Unknown
- Model catalog
- Unknown
- Monitoring
- Not stated
- Evaluation
- Not stated
Security & compliance
No controls are stated.
Geography
Provider headquarters are separate from offering infrastructure and availability.
Infrastructure regions
Global
Data-residency jurisdictions
Global
Service availability
Not stated
Support coverage
Not stated
Available plans
Pricing can change. Promotional prices are shown separately from standard prices.
| Plan | Price | Compute | Storage | GPU | Price checked | Actions |
|---|---|---|---|---|---|---|
| Amazon Bedrock, SageMaker AI & EC2 GPU InstancesNot stated | $0.04Usage Only | Unknown — RAM | Unknown | No | 7 October 2026 |
Data sources & provenance
Where this information came from, how it was checked and when it was last looked at. “Last updated” records a change to the record — it does not mean the record was verified.
- Data quality
- Data quality: Unverified
- Primary source
- aws.amazon.com
- Sources cited
- 2
- Last reviewed
- 7 October 2026
- Last updated
- 7 October 2026