Grok 4.7 from xAI is now available on Amazon Bedrock

Amazon Web Services has added xAI's Grok 4.7 to the Amazon Bedrock model catalog, the company wrote on its Machine Learning Blog. The model is positioned as a frontier system for coding, long-running agents and knowledge work, with a 500K token context window and four reasoning effort levels: low, medium, high and xhigh.

Grok 4.7 accepts text and image input and returns text. It runs on the bedrock-runtime endpoint through cross-Region inference profiles and supports the Responses, Chat Completions, InvokeModel and Converse interfaces.

What xAI says the model is built for

According to xAI's launch announcement of September 21, 2026, Grok 4.7 is the company's most capable model for coding and knowledge work. The theme is endurance rather than raw speed: it works longer on difficult tasks and verifies its own output more carefully before moving on.

The model builds on a new and larger base model, trained with a longer reinforcement learning run over a harder mix of tasks weighted toward problems that take many hours. That produced two capabilities, xAI says: better self-verification and more effective use of the 500K token context window on long tasks. xAI also reports stronger document and presentation generation.

Independent evaluation

Artificial Analysis, which runs its own evaluations rather than relying on developer-reported figures, measured Grok 4.7 at the xhigh effort. The Intelligence Index rose to 46 from 44, the Coding Agent Index to 56 from 47, AA-Briefcase Elo to 1,657 from 1,546, and GDPval-AA to 1,695 from 1,605.

The AA-Omniscience hallucination rate fell to 29% from 34%. One tradeoff stands out: output tokens per Intelligence Index task roughly doubled, from about 38,000 to about 81,000. That is why AWS advises setting the effort level deliberately instead of inheriting the default, which is high.

Packaging, safety and cost controls

Requests must name one of two cross-Region inference profiles. The global profile, global.xai.grok-4.7, routes each request to any supported commercial AWS Region and is priced below the geographic profile. The US profile, us.xai.grok-4.7, keeps processing inside the United States.

The model is OpenAI-compatible, so developers can reach it with the OpenAI SDK or through the AWS SDKs with Converse. Bedrock Guardrails apply content filters and PII redaction, structured outputs constrain responses to a JSON Schema, and invocation logging records calls in Amazon CloudWatch. According to xAI, Grok 4.7 ships with an entirely new safeguard stack.