Grok 4.7 is now available on Amazon Bedrock

xAI has released Grok 4.7 to Amazon Bedrock, introducing a powerful model designed for coding, long-running agents, and knowledge work. This model boasts a 500K token context window and offers four reasoning effort levels: low, medium, high, and xhigh. It is accessible via the bedrock-runtime endpoint through cross-Region inference profiles and supports Responses, Chat Completions, and Converse APIs.

According to xAI, Grok 4.7 is their most capable model for coding and knowledge work, working longer on complex tasks and verifying output more diligently before proceeding. The model was trained on a larger base model with a longer reinforcement learning run focused on difficult tasks, enhancing its ability to verify work and utilize its 500K token context window effectively.

Independent evaluations have shown improvements in long-horizon agentic knowledge work and coding agents, with the Intelligence Index increasing from 44 to 46. However, the output tokens per task nearly doubled, so users should set the effort level deliberately. Grok 4.7 comes with an improved safeguard stack, making it stronger against refusals and jailbreak resistance.