Grok 4.6 Reaches General Availability on Amazon Bedrock

Grok 4.6 became generally available on Amazon Bedrock on August 19, 2026, marking xAI's flagship model's second major cloud home just one week following its initial release. [1] The model is accessible via both of Bedrock’s endpoints: the standard bedrock-runtime endpoint and bedrock-mantle, the inference engine used for the first Grok release. [1] This deployment is significantly wider than the earlier Grok listing, with Grok 4.6 reachable through Geo and Global cross-Region inference profiles in over 30 AWS Regions on the standard endpoint. [1]

The model, listed under the ID `xai.grok-4.6`, supports text and image input with text-only output. [1] xAI configured the model with a 500,000-token context window and four reasoning effort settings: low, medium, high, and xhigh. [1]

Pricing varies based on the deployment method chosen by users. [1] Direct pricing from xAI is set at $2 per million input tokens and $6 per million output tokens. [1] On Bedrock, In-Region and Geo cross-Region requests cost $2.20 per million input tokens and $6.60 per million output tokens, while Global cross-Region inference is priced at $2.00 and $6.00 respectively. [1] AWS also offers cached-read pricing at $0.55 per million tokens for In-Region and Geo requests, and $0.50 for Global cross-Region inference. [1]

The rapid deployment of Grok 4.6 to Amazon Bedrock just a week after launch indicates xAI's intent to aggressively position its flagship model in major enterprise cloud environments. The tiered pricing structure on AWS, offering discounts for Global cross-Region routing and cached reads, provides cost-optimization levers for high-volume enterprise users. By supporting 30+ AWS Regions immediately, the deployment significantly outpaces the regional footprint of the earlier Grok 4.3 listing, suggesting improved infrastructure readiness. What the sources don't address: How the latency and performance characteristics of Grok 4.6 on Bedrock's two different endpoints (mantle vs. standard runtime) compare in production workloads.