xAI announced on August 19 that Grok 4.6 is generally available through Amazon Bedrock. The release gives developers in supported AWS regions access to the model through Amazon’s cloud platform, with xAI directing users to AWS documentation for setup instructions.
The company describes Grok 4.6 as its latest flagship model, designed for long-running agents as well as interactive and visual workloads. Its stated context window is 500,000 tokens, allowing a single request to accommodate a large amount of input, subject to the way developers configure and use the service.
Amazon Bedrock users can select among four configurable reasoning-effort levels: low, medium, high and xhigh. These settings provide different levels of reasoning effort for a request. The announcement does not include benchmark results or comparative figures for the speed, cost or output quality of the four modes.
Published pricing is $2 per million input tokens and $6 per million output tokens. Cached input is priced at $0.50 per million tokens. The announcement does not specify additional pricing conditions or describe whether rates vary among supported AWS regions.
xAI says Grok 4.6 is available to all developers in supported AWS regions. Its announcement does not provide a region-by-region availability list or reproduce the activation procedure, instead referring prospective users to Amazon Web Services documentation to begin building with the model.
Practical context: For teams already using AWS infrastructure, the Bedrock release provides another route for using Grok 4.6 without moving the workload to a separate model-provider platform. Cost calculations should distinguish among standard input, cached input and generated output: at the published rates, output tokens cost three times as much as ordinary input tokens and twelve times as much as cached input.
| Processing type | Price per 1 million tokens |
|---|---|
| Input | $2 |
| Cached input | $0.50 |
| Output | $6 |
Sources
Event date: 2026-08-19. Primary source date: 2026-08-19.
