Amazon Web Services has added SpaceXAI’s Grok 4.6 to its Bedrock platform, offering the frontier model with cross-region inference routing that automatically distributes requests across multiple AWS regions for higher throughput and reduced costs.
The integration, announced on August 19, gives AWS customers access to Grok 4.6’s 500,000-token context window, configurable reasoning efforts ranging from low to extra-high, and two distinct inference profiles designed to meet different data residency and performance needs.
Two Inference Profiles for Different Needs
AWS is offering two cross-region inference profiles for Grok 4.6. The US Geo profile, us.xai.grok-4.6, routes requests only within United States geography, allowing companies to scale their AI workloads while keeping all data processing within the country to meet regulatory requirements. The Global profile, global.xai.grok-4.6, serves requests from any commercial AWS region where the model is available, providing the broadest access to Bedrock capacity.
Cross-region inference is a feature that automatically distributes inference requests across multiple AWS regions to deliver higher throughput without requiring customers to manage capacity themselves. During demand spikes, the global profile offers the highest throughput at a lower per-token cost, according to AWS.
Competing in the Frontier Model Race
Grok 4.6, which launched on the xAI API in mid-August, is SpaceXAI’s latest frontier model built for coding, agentic tasks, and knowledge work. It features a 500K context window large enough to ingest entire codebases in a single pass, making it attractive for repository-scale development tasks.
The model joins a growing roster of frontier AI models available on Bedrock, which already hosts Anthropic’s Claude family, Meta’s Llama models, and Cohere’s offerings. By adding Grok 4.6, AWS is strengthening its position as a neutral platform where customers can access models from competing AI labs without being locked into a single provider.
Grok 4.6 supports the Bedrock Responses, Chat Completions, and Converse APIs, and integrates with existing account-level controls including model invocation logging, CloudWatch metrics, and cost itemization in AWS Cost Explorer. The model is available in all AWS regions where Bedrock is offered.
The launch comes amid intensifying competition among cloud providers to host frontier AI models. Google Cloud offers Gemini models and partners with Anthropic for Claude access, while Microsoft Azure is deeply integrated with OpenAI’s GPT family. AWS’s strategy of hosting models from multiple competing labs, including SpaceXAI’s Grok, positions it as the cloud provider most focused on giving customers model choice.
Sources: AWS announcement; xAI release notes; AI Weekly
discussion