Added Claude Fable 5.1 and Claude Mythos 5.1: Added model cards for Claude Fable 5.1 and Claude Mythos 5.1 . Updated the Adaptive thinking, Extended thinking, Tool use, and Prompt caching pages. Added a new Thinking block binding (beta) topic.

Anthropic Models Available in Amazon Bedrock

The following Anthropic models are now available:

• Claude Fable 5.1 - Frontier model for coding, long-horizon agents, and enterprise knowledge work.
• Claude Mythos 5.1 - Capable model for cybersecurity defense and life sciences research.
• Claude Opus 5 - Advanced Opus model for long-running agents and coding.
• Claude Sonnet 5 - Capable Sonnet model for coding, agents, and professional work at scale.
• Claude Mythos 5 - Model for cybersecurity and life sciences, including vulnerability discovery and drug design.
• Claude Fable 5 - Next-generation model for complex knowledge work and coding.
• Claude Opus 4.8 - Optimized for coding, agents, and deeper reasoning in enterprise workflows.
• Claude Opus 4.7 - Built for coding, enterprise workflows, and long-running agentic tasks.
• Claude Mythos Preview - Gated research preview for ambitious projects and the world's best model for cybersecurity, autonomous coding, and long-running agents.
• Claude Sonnet 4.6 - Mid-tier model with improved coding, long-context reasoning, and agent planning.
• Claude Opus 4.6 - Flagship model with improved planning, sustained agentic tasks, and reliability in massive codebases.
• Claude Opus 4.5 - Model for coding, agents, and computer use with improvements for spreadsheets and long-running chats.
• Claude Haiku 4.5 - Lightweight model optimized for speed and efficiency with strong coding and agent performance.
• Claude Sonnet 4.5 - Optimized for agents, coding, and computer use with significant improvements across all benchmarks.
• Claude Opus 4.1 - Upgrade with improved coding, reasoning, and agentic task capabilities.
• Claude Sonnet 4 - Balanced model with strong coding and reasoning capabilities, improved instruction following, and extended thinking with tool use.
• Claude 3.5 Haiku - Next-generation fast model with improved coding and reasoning performance.
• Claude 3 Haiku - Fastest and most compact model, optimized for speed and efficiency.

Billing Notes

Refused requests (all Anthropic models on bedrock-mantle):

• No input or output tokens are billed when a request is refused by a content classifier before inference begins.
• Mid-stream refusals — where the classifier fires after output has begun streaming — are billed for tokens consumed before the block.
• Both refusal types return stop_reason: "refusal" in the response.

Source: AWS release notes

If you need further guidance on AWS, our experts are available at AWS@westloop.io. You may also reach us by submitting the Contact Us form.