--- title: Amazon Bedrock description: Fully managed service providing access to foundation models from Amazon, Anthropic, Meta, Mistral, and others — for building generative AI applications. url: https://www.joinhorizon.co/glossary/amazon-bedrock/ publishDate: 2026-10-08 updateDate: 2026-10-08 --- # Amazon Bedrock > Fully managed service providing access to foundation models from Amazon, Anthropic, Meta, Mistral, and others — for building generative AI applications. > **AWS lifecycle notice (June 30, 2026)** — Amazon Bedrock Agents is now **Bedrock Agents Classic**, in maintenance for new customers after **July 30, 2026**. Net-new agent builds should use [Bedrock AgentCore](https://www.joinhorizon.co/blog/amazon-bedrock-agentcore-production/). Full matrix: [lifecycle roundup](https://www.joinhorizon.co/blog/aws-service-lifecycle-updates-june-2026/). ## Definition Amazon Bedrock is a fully managed generative AI service that exposes foundation models (FMs) through a unified API inside your AWS account. As of October 2026, Bedrock offers **roughly 100 serverless models** from Amazon, Anthropic, Meta, Mistral, Cohere, OpenAI, xAI, and others — including **Claude Fable 5.1** (September 1, 2026 — Covered Model; 1M context), **Claude Opus 5.5** (September 22, 2026 — ZDR default on Bedrock; adaptive thinking always on; 1M context), **Claude Sonnet 5.5** (September 28, 2026 — 1M context; Global CRIS `global.anthropic.claude-sonnet-5-5`; also in AWS GovCloud (US)), **Claude Sonnet 5** (June 30, 2026 — prior Sonnet pin), **Claude Opus 5** (July 24, 2026 — prior Opus pin), **Claude Haiku 5.5** (October 7, 2026 — current Haiku, 1M context, model ID `anthropic.claude-haiku-5-5`, Global CRIS `global.anthropic.claude-haiku-5-5`; **Claude Haiku 4.5**, October 16, 2025, remains the prior pin), **GPT-6 Astra** (September 8, 2026 — hardest end-to-end OpenAI model; 1,050,000-token context), **GPT-6 Sol and Luna** (September 22, 2026 — daily complex work and high-volume tasks; up to 1M context; explicit prompt caching), **GPT-5.6 Sol, Terra, and Luna** (July 2026 — prior OpenAI pins; **1M token** context as of August 3, 2026), **Grok 4.7** (September 28, 2026 — 500K context; reasoning effort low/medium/high/xhigh; US Geo `us.xai.grok-4.7` and Global `global.xai.grok-4.7` CRIS on `bedrock-runtime`), **Grok 4.6** (August 19, 2026 — prior xAI pin), **Grok 4.3**, **Fable 5** (June 9, 2026 — prior Mythos-class pin), **Opus 4.8** (May 28, 2026), **Sonnet 4.6**, the **Amazon Nova** family (**Nova 2 Lite** for everyday multimodal with 1M context, plus Nova 1 Micro through Premier, **Canvas**, **Reel**, and **Nova 2 Sonic**), marketplace pins **GLM 5.3** (October 5, 2026 — eligible customers, US CRIS `us.zai.glm-5.3` or Global CRIS `global.zai.glm-5.3`, 1M context, 744B total / ~40B active; **GLM 5**, March 18, 2026, remains the prior pin) and **MiniMax M2.5** (March 18, 2026), **Mistral Large 3**, and open-weight options such as Llama. Bedrock also provides **Knowledge Bases** (managed RAG with **S3 Vectors** or OpenSearch backends), **Agents Classic** (maintenance for new customers after July 30, 2026), **Guardrails**, **Prompt Flows**, **Data Automation**, and the **bedrock-mantle** endpoint (OpenAI Responses/Chat Completions and Anthropic Messages APIs). Net-new multi-step agents should use [Bedrock AgentCore](https://www.joinhorizon.co/glossary/bedrock-agentcore/). Model inference and customer content stay within your AWS environment under the standard AWS shared responsibility model. Covered Models (Fable 5 / 5.1) require Data Retention API mode `aws_review` and 30-day minimum retention; they are not ZDR by default. ## When to use it - **Production GenAI without GPU fleet management** — pay per token instead of operating inference clusters. - **Model diversity in one integration** — route classification and everyday multimodal to **Nova 2 Lite** (or Nova Micro for cheapest text), high-volume chat to **GPT-6 Luna** or **Claude Haiku 5.5** (Haiku 4.5 remains the prior Haiku pin; GPT-5.6 Luna remains the prior OpenAI volume pin), recurring complex work and coding on OpenAI to **GPT-6 Sol**, the hardest OpenAI end-to-end jobs to **GPT-6 Astra**, agentic coding on Claude to **Claude Sonnet 5.5** (Sonnet 5 remains the prior pin), maximum Opus-class reasoning (with ZDR default) to **Claude Opus 5.5** (Opus 5 remains the prior pin until you benchmark), long-horizon autonomous jobs to Claude Fable 5.1 after Covered Model legal sign-off, and **Grok 4.7** when you need Converse-native xAI plus US Geo / Global CRIS from the same codebase (Converse API or bedrock-mantle). - **Grounded Q&A over private documents** via Knowledge Bases and the Converse / RetrieveAndGenerate APIs. - **Regulated workloads** requiring Guardrails, CloudTrail logging, VPC endpoints, and IAM-scoped model access. - **Multi-step agents** that call APIs, query knowledge bases, and retain session memory — extend with [Bedrock AgentCore](https://www.joinhorizon.co/glossary/bedrock-agentcore/) for managed runtime infrastructure. ## When not to use it - **Lowest cost at very high sustained throughput** on a single fixed model — self-hosted inference on SageMaker or EC2 can win economically after careful modeling. - **Fully custom architecture not available on Bedrock** — exotic fine-tuning, non-supported frameworks, or on-device inference need different stacks. - **Single-turn tasks solvable without an LLM** — traditional search, rules engines, or classical ML may be cheaper and more deterministic. ## Tips - Enable **model access** per region for each provider in the Bedrock console before deployment — missing access shows as obscure API errors in CI. - Apply **Guardrails** in production on both input and output regardless of model vendor. - Run **model evaluation** on your own prompt set before standardizing on Opus-class models — Sonnet 5.5, Haiku 5.5 (Haiku 4.5 remains the prior pin), or Nova 2 Lite often meet the quality bar at lower cost. - Use **cross-region inference profiles** where available for resilience when a single region throttles. - Separate **dev and prod** IAM policies and Knowledge Base data sources to prevent test prompts from retrieving production documents. ## Gotchas ### Serious - **No Guardrails on customer-facing endpoints** — models can leak training-adjacent patterns, generate harmful content, or echo PII from prompts. - **Throttling during launch** — new workloads need quota headroom and exponential backoff; cross-region profiles help but do not eliminate limits. - **Treating all models as interchangeable** — tool use, context length, multimodal input, and fine-tuning support vary by model ID. ### Regular - **Hardcoding model IDs** — AWS releases new versions; pin in config and plan upgrade testing when IDs deprecate. - **Skipping citation review in RAG** — Knowledge Bases reduce hallucination but do not eliminate wrong-chunk retrieval; validate sources in UI. - **Logging full prompts with secrets** — CloudWatch logs may capture API payloads; redact credentials and PHI at the application layer. ## Official references - [What is Amazon Bedrock?](https://docs.aws.amazon.com/bedrock/latest/userguide/what-is-bedrock.html) - [Supported foundation models](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html) - [Amazon Bedrock Knowledge Bases](https://docs.aws.amazon.com/bedrock/latest/userguide/knowledge-base.html) - [Guardrails for Amazon Bedrock](https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails.html) ## Related FactualMinds content - [Generative AI on AWS Bedrock](https://www.joinhorizon.co/services/generative-ai-on-aws/) - [AWS Bedrock Consulting](https://www.joinhorizon.co/services/aws-bedrock/) - [Fine-Tuning vs RAG on AWS Bedrock: When to Use Each](https://www.joinhorizon.co/blog/fine-tuning-vs-rag-bedrock-when-to-use/) - [RAG Pipeline](https://www.joinhorizon.co/glossary/rag-pipeline/) ## Related AWS Services - generative-ai-on-aws - aws-bedrock ## Related Posts - claude-opus-5-5-aws-bedrock-2026 - openai-gpt-6-sol-luna-amazon-bedrock-2026 - fine-tuning-vs-rag-bedrock-when-to-use - claude-fable-5-aws-bedrock-mythos-class-guide-2026 --- *Source: https://www.joinhorizon.co/glossary/amazon-bedrock/*