xAI Grok 4.6 goes live on Amazon Bedrock, enters AWS enterprise model distribution
On August 19 xAI announced Grok 4.6 is generally available on Amazon Bedrock with US Geo and Global cross-region inference at $2/$6 per 1M tokens.
On August 19, xAI announced Grok 4.6 is generally available on Amazon Bedrock, letting AWS customers invoke the model via Responses, Chat Completions, or Converse APIs. This is Grok's first entry into the AWS enterprise model distribution system after its own API, Cursor, Grok Build, and GitHub Copilot.
Grok 4.6 retains its 500K token context window on Bedrock with low, medium, high, and xhigh reasoning effort levels. Bedrock offers two inference profiles: US Geo (us.xai.grok-4.6) which restricts processing to US regions for data residency, and Global (global.xai.grok-4.6) which routes across all Bedrock-available regions for higher throughput and lower per-token cost during peaks. Pricing aligns with xAI's direct API at $2 input and $6 output per 1M tokens.
Bedrock provides model invocation logging (deliverable to Amazon S3 or CloudWatch Logs), CloudWatch metrics, and cost breakdowns in Amazon Cost Explorer and Cost and Usage Reports, so enterprise IT can manage Grok 4.6 under existing IAM, logging, and billing. xAI explicitly states it does not use customer data for training.
Industry observers note that with Grok on AWS Bedrock, model distribution has now expanded from API, Cursor, and Copilot into enterprise cloud IAM, logging, cost, and audit systems. Grok 4.7 remains a teaser with no model card, API identifier, or third-party benchmarks yet.