What's New

Amazon Bedrock has launched a redesigned console experience purpose-built for the bedrock-mantle endpoint, which supports the OpenAI Responses API, OpenAI Chat Completions API, and the Anthropic Messages API. The new console organizes work into projects and streamlines the full lifecycle from model discovery and evaluation to production deployment. It introduces project-aware documentation that auto-populates code snippets with the correct model ID, region, endpoint URL, and API key, enabling developers to copy and run code without modification.

How It Works

• bedrock-mantle endpoint: The new console is optimized for the bedrock-mantle.{region}.api.aws endpoint, which natively supports OpenAI-compatible APIs (Responses and Chat Completions) and the Anthropic Messages API, allowing existing OpenAI or Anthropic SDK codebases to connect by changing only the base URL and API key.
• Project-based organization: Work is structured into projects that group model selection, evaluations, and usage insights together, mirroring the real-world lifecycle of building a generative AI application.
• Model catalog and comparison: Customers can browse the full Bedrock model catalog—including Claude, GPT, and open-weight models—and compare them side by side on capabilities, modality support, context window size, and service quotas in a single unified view.
• Project-aware documentation: Code samples, SDK snippets, and API references are automatically prefilled with the project's selected model ID, region, endpoint URL, and API key reference, and update dynamically as settings change.
• Authentication via Bedrock API key: The bedrock-mantle endpoint supports Bedrock-native API keys (in addition to AWS SigV4), making it directly compatible with the OpenAI SDK's authentication model.
• Built-in capabilities: The endpoint includes stateful conversation management, server-side tool use, and advanced scheduling/work-queueing for higher initial throughput compared to the per-account quota model of bedrock-runtime.

Why It's Important

• Reduces migration friction: Developers already using OpenAI or Anthropic client libraries can switch to Amazon Bedrock by changing only the base URL and API key, dramatically lowering the barrier to adoption.
• Eliminates documentation sprawl: The side-by-side model comparison and auto-populated code snippets remove the need to cross-reference multiple documentation pages, limit calculators, and pricing sheets.
• Accelerates time-to-production: The project-based workflow integrates evaluation, iteration, and deployment insights into one place, shortening the typical experimentation-to-production cycle.
• Improves developer experience: Copy-paste-ready code snippets that require zero modification reduce errors and onboarding time for new team members or projects.
• Signals AWS's competitive positioning: By explicitly surfacing GPT models alongside Claude and open-weight models in the same catalog, AWS is positioning Bedrock as a neutral, multi-model platform competing directly with OpenAI's and Anthropic's own developer consoles.
• Enterprise-grade throughput model: The bedrock-mantle endpoint's advanced scheduling architecture supports higher initial throughput than the fixed-quota bedrock-runtime model, making it more suitable for bursty or rapidly scaling workloads.

How It's Different

• vs. the existing bedrock-runtime console: The legacy console is optimized for Bedrock-native APIs (InvokeModel, Converse) with per-account fixed throughput quotas, whereas the new experience targets the bedrock-mantle endpoint with OpenAI/Anthropic API compatibility and a work-queue-based throughput model.
• vs. OpenAI's developer console: Bedrock's new console adds enterprise features like IAM integration, per-project usage attribution, cross-region inference profiles, and AWS-native security controls that OpenAI's platform does not offer.
• vs. Anthropic's console: While Anthropic's console is limited to Claude models, the Bedrock console surfaces Claude, GPT, and open-weight models in a single catalog with unified quota and capability comparisons.
• Project-aware dynamic documentation: Unlike static documentation or generic quickstart guides, the console generates code snippets that are pre-configured for the specific project context and update in real time as the user changes models or regions.
• Unified evaluation and insights: Competitors typically separate model evaluation tools from production monitoring; Bedrock's project model integrates both into the same workflow.
• Dual-endpoint strategy: AWS maintains both bedrock-mantle (for OpenAI/Anthropic compatibility) and bedrock-runtime (for native Bedrock APIs), giving customers flexibility rather than forcing a migration.

When to Prefer It

• Migrating from OpenAI or Anthropic: Use the new console and bedrock-mantle endpoint when porting an existing application built on the OpenAI SDK or Anthropic SDK, as only the base URL and API key need to change.
• Rapid prototyping and model comparison: Use it when you need to quickly evaluate multiple models (Claude, GPT, open-weight) side by side on capabilities and quotas before committing to one for production.
• Team-based or multi-project development: Use it when multiple developers or teams are working on distinct generative AI applications and need isolated project contexts with separate usage tracking.
• Stateful conversational applications: Prefer bedrock-mantle when building chatbots or multi-turn agents that benefit from built-in stateful conversation management rather than managing conversation history client-side.
• Server-side tool use: Use it when your application requires server-side tool execution, which is supported on bedrock-mantle but not on bedrock-runtime.
• Bursty or high-growth workloads: Prefer bedrock-mantle when your traffic pattern is unpredictable or rapidly scaling, as its work-queue architecture supports higher initial throughput than the fixed-quota model of bedrock-runtime.
• Stick with bedrock-runtime when: You are using Bedrock-native InvokeModel or Converse APIs, need cross-region inference profiles, or require a model not yet available on the bedrock-mantle endpoint.

Availability

• Status: Generally available (GA) as of June 4, 2026; this is a console and endpoint experience update to the existing Amazon Bedrock service, not a preview.
• Supported regions: US East (N. Virginia, Ohio), US West (Oregon), Asia Pacific (Jakarta, Mumbai, Sydney, Tokyo), Europe (Frankfurt, Ireland, London, Milan, Stockholm), and South America (São Paulo)—all regions where the bedrock-mantle endpoint is offered.
• Access: Available via the AWS Management Console at console.aws.amazon.com/bedrock-mantle/; select the new experience from the Bedrock navigation pane.
• Pricing: No separate charge for the console experience itself; standard Amazon Bedrock inference pricing applies based on the model selected and tokens consumed through the bedrock-mantle endpoint.
• Authentication: Supports both Bedrock API keys (compatible with OpenAI SDK) and AWS SigV4; IAM-based access control and usage attribution are available.
• Limitations: Not all models available on bedrock-runtime are currently available on bedrock-mantle; check the Endpoint availability documentation for per-model support. The bedrock-runtime endpoint remains fully supported for native Bedrock API use cases.

Related Resources

• https://docs.aws.amazon.com/bedrock/latest/userguide/endpoints.html
• https://console.aws.amazon.com/bedrock-mantle/