Grok 4.6 Lands on Microsoft Foundry

SpaceXAI has made Grok 4.6 available on Microsoft Azure AI Foundry, giving Azure customers an official, managed path to the company’s latest flagship model inside Microsoft’s enterprise model catalog.

SpaceXAI confirmed the availability of Grok 4.6 via Microsoft Foundry in a post on their website. Microsoft’s Foundry is a place for organizations to evaluate various frontier models, including Grok 4.6, run workload-specific tests, deploy managed endpoints, and operate with enterprise-grade security and governance controls.

The Microsoft Foundry catalog now lists Grok 4.6, alongside older versions like Grok 4.3, Grok 4.20 Reasoning, and Grok 4.1 Code Fast. The Foundry Integration Guide has also been updated to include support for Grok 4.6, including the basic steps to deploy it to Azure AI.

Public Preview

Microsoft’s Foundry blog confirmed that Grok 4.6 is now available as a public preview. According to that post, Grok 4.6 from SpaceXAI is available in Microsoft Foundry Models through public preview on a unified platform for model discovery, evaluation, deployment, and governance. Microsoft said organizations can assess Grok 4.6 against their own workloads, business requirements, and production constraints before selecting and deploying a configuration.

SpaceXAI launched Grok 4.6 just under two weeks ago, and the latest model builds on the already competent Grok 4.5 with a focus on long-running agentic tasks, alongside more interactive and visual work capabilities. That all comes with a 500,000-token context window and configurable reasoning efforts for the model, enabling additional granularity for users and developers.

Currently, on the Azure AI Foundry, Grok 4.6 comes with low, medium, high, and xhigh as reasoning parameters.

Enterprise Deployments

The Microsoft listing closes a fast cloud-and-partner push. After the August 12 release, SpaceXAI’s news index shows Grok 4.6 reaching GitHub Copilot on August 14, Amazon Bedrock, and Google’s Gemini Enterprise Agent Platform before the August 26 Foundry post. That means the flagship Grok 4.6 model is now available through most managed enterprise deployment paths in North America.

For enterprise buyers who already standardize on Azure AI Forge identity, networking, and compliance tooling, this is a big leap over the previously available Grok 4.3. For companies looking to deploy against the latest frontier models, this now means they can finally deploy Grok against other models like GPT 5.6 Sol, Claude Opus/Fable 5, and Kimi K2.7 Code.

Enterprise API Pricing

API pricing for Grok 4.6, from SpaceXAI’s August 12 announcement, starts at $2 per million input tokens and $6 per million output tokens, with a faster variant listed at twice that price.

For those not familiar with enterprise frontier model pricing, this may seem quite expensive, but compared against GPT 5.6 Sol and Claude Opus 5, Grok 4.6 offers relatively cheap and competitive pricing options.