xAI’s Grok 4.6 Now Available on Amazon Bedrock
xAI’s Grok 4.6, a model designed for long-running agents, coding, and knowledge work, is now available on Amazon Bedrock with a 500K token context window.
Today, AWS announced the availability of xAI’s Grok 4.6 on Amazon Bedrock, adding a new frontier model to the Bedrock model catalog. Grok 4.6, which launched on Bedrock on August 18, 2026, offers a 500K token context window and supports configurable reasoning effort at four levels: low, medium, high, and xhigh. This is xAI’s second model in Amazon Bedrock, following the general availability of Grok 4.3 in Bedrock Mantle, an OpenAI-compatible inference engine in Amazon Bedrock. Grok 4.6 is now available on both the bedrock-mantle and bedrock-runtime endpoints, and supports the Converse API alongside Chat Completions and Responses.
xAI reports that Grok 4.6 builds on Grok 4.5 with a focus on long-running agents and more ambitious interactive and visual work. The model is designed to handle complex tasks across many steps, such as researching a topic, analyzing information, working across a code base, or turning an idea into a polished application or work artifact. Training details include a longer supplemental training run using curated model-generated data for reasoning and advanced technical concepts, high-quality engineering data, and an improved optimizer and training recipe. The model was then trained on a wide range of agentic reinforcement learning tasks spanning knowledge work, general coding, and domain-specific environments.
Grok 4.6 is served on bedrock-runtime in addition to bedrock-mantle, allowing access through the AWS SDKs and the standard Bedrock control surface. The Converse API, including streaming, is also available. An xhigh reasoning effort level is supported, and cross-region inference is enabled with two inference profiles: us.xai.grok-4.6 and global.xai.grok-4.6. The global profile is cheaper at $2.00 per million input tokens compared to $2.20 for the US profile, making it the default choice unless data residency requirements dictate otherwise.
Source: aws-ml

