AWS Adds xAI’s Grok 4.7 to Amazon Bedrock With 500K Context Window

Amazon Web Services announced on September 28, 2026, that xAI’s Grok 4.7 model is available on Amazon Bedrock, adding it to the Bedrock model catalog with a 500K token context window, text and image input, and four configurable reasoning effort levels. The AWS Machine Learning Blog announcement describes Grok 4.7 as a frontier model built for coding, long-running agents, and knowledge work.

xAI first released Grok 4.7 on September 21, 2026, in a launch announcement that made the model available in Cursor and Grok Build and through the Grok API, third-party coding harnesses, and model routers and cloud platforms. In that announcement, xAI priced the model starting at $2 per million input tokens and $6 per million output tokens, and said it serves a fast variant at twice the output speed for twice the price.

What xAI Says Grok 4.7 Is Built For

According to xAI, as cited in the AWS post, Grok 4.7 is its most capable model for coding and knowledge work, with an emphasis on endurance rather than raw speed: the model works longer on difficult tasks and checks its own work more carefully before moving on. xAI reports that Grok 4.7 uses a new, larger base model trained through a longer reinforcement learning run over a harder mix of tasks, weighted toward problems that take many hours to complete. xAI credits that training with two capabilities: better verification of the model’s own work, and more effective use of its 500K token context window on long tasks. xAI also says it trained the model to natively understand the Grok Bot harness, which it credits for improvements in conversational tasks and general knowledge work.

xAI further reports stronger document and presentation generation and describes gains on professional knowledge work of the kind done by lawyers, nurses, and financial analysts. Its published evaluations span software engineering with CursorBench and DeepSWE, multi-hour terminal and office work with Terminal-Bench and AA Briefcase, electrical engineering with EEBench, legal work with the Harvey Legal Agent Benchmark, and clinical reasoning with HealthBench Professional.

Independent Evaluation Figures

The AWS post also cites Artificial Analysis, which it describes as running its own evaluations rather than relying on developer-reported figures. In those results, Grok 4.7 records an Intelligence Index of 46 against 44 for Grok 4.6, a Coding Agent Index of 56 against 47, an AA-Briefcase Elo of 1,657 against 1,546, a GDPval-AA Elo of 1,695 against 1,605, and an AA-Omniscience Index of 32 against 30, with a hallucination rate of 29% against 34%. Artificial Analysis measured Grok 4.7 at its xhigh reasoning effort and Grok 4.6 at the effort level it reported for each measure.

The post flags the tradeoff in the final row of that table: Grok 4.7 used roughly 81,000 output tokens per Intelligence Index task, against about 38,000 for Grok 4.6, which the post characterizes as roughly double per task and a reason to set the effort level deliberately rather than inherit the default.

Safety and Cyber Security

According to xAI, Grok 4.7 was built with an entirely new safeguard stack and is the strongest model it has tested on refusals and jailbreak resistance. xAI frames the goal in dual-use domains such as cyber security and biological work as remaining useful for legitimate tasks while refusing dangerous ones. On cyber security specifically, xAI reports the model allows only a small fraction of risky dual-use prompts through while rarely blocking legitimate security work, and it has begun giving selected cyber security partners invite-only access to Grok 4.7’s red-team capabilities for defense research.

How Grok 4.7 Is Packaged on Amazon Bedrock

On Bedrock, Grok 4.7 accepts text and image input and returns text. The model is served on the bedrock-runtime endpoint through cross-Region inference profiles, so requests name a profile rather than a bare model ID: us.xai.grok-4.7 for the US geographic profile, or global.xai.grok-4.7 for the Global profile, both at the /openai/v1 path of the bedrock-runtime base URL. Grok 4.7 supports the Responses API, the Chat Completions API, InvokeModel, and the Converse API.

Because the model is OpenAI-compatible, the OpenAI SDK works against the /openai/v1 path with a bearer token, which can be either an Amazon Bedrock API key or a short-term token minted from AWS Identity and Access Management (IAM) credentials. The AWS SDKs reach the same model through Converse, signing requests with ordinary AWS credentials.

The model also connects to standard Bedrock features. Implicit prompt caching applies automatically to repeated prompt prefixes, so agents that resend a large system prompt or reference document on every turn pay the cached rate for that prefix. Bedrock Guardrails attach by ID and version on the request, applying content filters, denied topics, personally identifiable information redaction, and word policies to both the prompt and the response. Structured outputs constrain a response to a JSON Schema so downstream code can parse it directly, and invocation logging captures each call in Amazon CloudWatch with the request, the response, and token counts including reasoning tokens.

Routing Profiles, Service Tiers, and Access

The Global profile routes each request to any supported commercial AWS Region, spreading load across more capacity, and is priced below a geographic profile. The post notes the tradeoff is less control over where a given request is served, which can mean more variable latency. The US geographic profile keeps processing within the US geography, addressing US data residency requirements.

Three service tiers are offered. Standard is pay-per-token with no commitment; Priority delivers faster, prioritized processing for a premium; and Flex provides lower-cost access for work that is not time-sensitive. Per-token pricing across the tiers appears on the Amazon Bedrock pricing page.

Before a first call, users confirm that the model is available to them in the Bedrock console for the AWS Region they plan to use. On permissions, bedrock:InvokeModel is evaluated against three resources: the account’s default project, the named inference profile, and the underlying foundation model, whose Amazon Resource Name is wildcarded across Regions because cross-Region profiles route outside the calling Region. Bearer-token authentication additionally requires bedrock:CallWithBearerToken, and profiles are scoped individually, so a policy naming the US profile does not cover the Global one.

Reasoning is always active on Grok 4.7. Effort is set through the reasoning parameter on the Responses API or through additionalModelRequestFields on Converse, with high as the default. Reasoning content is encrypted; it can be returned by passing include: “reasoning.encrypted_content” on a Responses API request and then sent back on subsequent turns to give the model its own prior reasoning as context in a multi-turn conversation. The Chat Completions API does not return reasoning tokens.

The post advises treating a long-term Amazon Bedrock API key as an exploration-only credential, using short-term bearer tokens generated from IAM credentials for production, and deleting an exploration key from the Bedrock console when it is no longer needed.