xAI's Grok 4.6 Launches on Amazon Bedrock With 500K Context and Cross-Region Inference

TL;DR

xAI's Grok 4.6 is now available on Amazon Bedrock via both bedrock-mantle and bedrock-runtime endpoints, adding Converse API support, cross-Region inference profiles, and Bedrock Guardrails. The model offers a 500K token context window and four reasoning effort levels, with input pricing starting at $2.00 per million tokens on the global inference profile.

3 min read
0

Grok 4.6 Reaches Amazon Bedrock

xAI's Grok 4.6 became available on Amazon Bedrock on August 18, 2026, six days after xAI launched the model itself on August 12, 2026. It is xAI's second model on the platform, following Grok 4.3, and it substantially widens the ways developers can access xAI models through AWS.

Grok 4.6 offers a 500K token context window and supports four configurable reasoning effort levels: low, medium, high, and xhigh — the last being new. xAI positions the model for long-running agentic work: multi-step research, codebase-wide analysis, and turning ideas into functioning applications or polished artifacts.

What's New on Bedrock

Unlike Grok 4.3, which was reachable only through Bedrock Mantle (Amazon Bedrock's OpenAI-compatible inference engine), Grok 4.6 is served on both bedrock-mantle and bedrock-runtime endpoints. That runtime support brings several capabilities:

  • Converse API, including streaming via converse and converse_stream, giving developers a consistent message format across Bedrock models instead of an OpenAI-compatible client only.
  • Cross-Region inference profiles: us.xai.grok-4.6 keeps traffic within the US for data residency needs, while global.xai.grok-4.6 routes worldwide for broader capacity. Global pricing is $2.00 per million input tokens versus $2.20 for the US profile — pricing not yet disclosed for output tokens in the source material.
  • Amazon Bedrock Guardrails support on bedrock-runtime, adding content filters, denied topics, PII redaction, and word policies evaluated against both prompts and model responses.
  • Model invocation logging to CloudWatch, capturing full request/response bodies, token counts (including reasoning tokens), and which inference profile handled the call.
  • Prompt caching, billed at roughly a quarter of standard input rates for repeated prefix content — relevant for agents that resend large system prompts on every turn.

The model accepts text and image input and returns text only; it does not support audio, video, or embeddings, and cannot generate images.

Benchmark Claims

According to xAI, Grok 4.6 High scored 61 on the Artificial Analysis Intelligence Index v4.1.1, a composite of nine evaluations including GDPval-AA v2, Humanity's Last Exam, GPQA Diamond, and AA-Omniscience. xAI also reported 69.9% on CursorBench v3.2, 65.9% on DeepSWE v1.1, 61.3% on FrontierCode v1.1 (Extended), 57.5% on APEX-Agents, 56.4% on APEX-SWE, 26% on Terminal-Bench v3.0, 1753 on GDPVal-AA v2, 1577 on AA-Briefcase, and 15.8% on Harvey LAB (Vals). These figures come from xAI's own launch announcement and have not been independently verified.

xAI states it used a longer supplemental training run than Grok 4.5, incorporating curated model-generated data, an improved optimizer, and agentic reinforcement learning across coding, kernel optimization, web development, and CAD tasks. The company also claims its widest-ever suite of pre-deployment safety testing for this release, though these are company claims rather than independently confirmed results.

What This Means

This is a distribution and packaging update, not a new model launch — Grok 4.6 itself debuted through xAI's own channels on August 12, 2026, six days before Bedrock availability. The news here is about access: enterprise AWS customers now get a second xAI model with standard Bedrock tooling (Converse API, Guardrails, CloudWatch logging, cross-Region routing) rather than being confined to an OpenAI-compatible shim. For teams already standardized on Bedrock's runtime and governance tooling, this lowers the friction of adopting xAI models without requiring a separate client integration. The $0.20 pricing gap between global and US-only inference profiles is a small but real signal that data residency requirements carry a measurable cost premium on this platform.

Related Articles

product update

Zed v1.30.0-pre adds GPT-6.1 Terra support and a commit message model picker

Zed's v1.30.0-pre release adds GPT-6.1 Terra to the OpenAI provider and a model picker for AI commit message generation in the Git Panel. It also adds an Extra High thinking effort level for Gemini 4 Hydrogen and reduces GPU usage across the app.

product update

Zed v1.29.1 adds Claude Fable 5.2 for Anthropic BYOK users and a Git Panel branch mode

Zed v1.29.1 adds support for Claude Fable 5.2 for Anthropic BYOK users, plus Claude Mythos 5.2 for organizations with access. The release also adds a Git Panel branch mode, a bookmark picker, configurable Git date formats and a setting to turn off automatic agent thread titles.

product update

Replit brings Microsoft Fabric, Teams and local work to general availability; Azure publishing in early access

At Microsoft Ignite, Replit announced general availability for its Microsoft Fabric integration, local access in Replit Desktop, and a Microsoft Teams integration. Publishing apps to a company's Azure tenant is in early access, with Full BYOC planned next. Replit was also named 2026 Microsoft Partner of the Year for Startups.

product update

GitHub adds eight more GitHub-verified MCP connectors to Copilot in public preview

GitHub says you can now connect eight more services to Copilot using GitHub-verified MCP servers, also called connectors. The connectors are in public preview. The announcement names Supabase, ClickUp and Buildkite, and lists Box for searching and managing content.

Comments

Loading...

Grok 4.6 Now Available on Amazon Bedrock | TPS