Superpower Daily
xAI Puts Grok 4.6 on Microsoft Foundry, Extending a Two-Week Cl...London Surgeons Remove 11mm Brain Tumour With Live AI Anatomy M...Chronograph Connects Perplexity to Private-Capital Data, Adding...Amazon’s Las Vegas Warehouse Reportedly Cuts Up Books for AI Tr...Lam Breaks Ground on Oregon Lab, First Step in $3B AI-Chip R&D...Hugging Face Explains DeepSeek’s Matched Shift From MoE Experts...xAI Puts Grok 4.6 on Microsoft Foundry, Extending a Two-Week Cl...London Surgeons Remove 11mm Brain Tumour With Live AI Anatomy M...Chronograph Connects Perplexity to Private-Capital Data, Adding...Amazon’s Las Vegas Warehouse Reportedly Cuts Up Books for AI Tr...Lam Breaks Ground on Oregon Lab, First Step in $3B AI-Chip R&D...Hugging Face Explains DeepSeek’s Matched Shift From MoE Experts...
Modelspublished

xAI Puts Grok 4.6 on Microsoft Foundry, Extending a Two-Week Cloud Rollout

Azure customers can now evaluate xAI’s model through managed endpoints and governance controls, but production terms for the public preview remain undefined.

By 2 min read
xAI Puts Grok 4.6 on Microsoft Foundry, Extending a Two-Week Cloud Rollout
Sign in to save and react

Listen to this story

The audio brief

About 1:34
0:001:34
Read transcript
Microsoft Foundry is adding xAI’s Grok 4.6, giving Azure customers a managed way to test the model alongside other leading systems. The release is a public preview, so teams can use managed endpoints and Microsoft’s security and governance controls, but it is not yet a statement of production availability or final commercial terms. The appeal is mainly distribution and comparison. Foundry lets customers evaluate models on their own data and choose deployment options in an existing cloud environment. Grok now appears beside offerings from OpenAI, Anthropic, Meta, Mistral, DeepSeek, Cohere, and Hugging Face. That makes this a new enterprise testing route, not evidence that customers have adopted the model. Grok 4.6 has already expanded beyond xAI’s API through Cursor, Grok Build, GitHub Copilot, Amazon Bedrock, and Google’s Gemini Enterprise Agent Platform. xAI positions it for sustained agent work, including navigating codebases, researching unfamiliar topics, building applications, and checking its own output. The model supports a context window of 500,000 tokens, with reasoning settings ranging from low to high, plus an extra-high, or “xhigh,” mode. xAI lists API pricing at two dollars per million input tokens and six dollars per million output tokens, but those numbers do not define Foundry pricing. Microsoft’s cited benchmarks are provider-supplied, with no independent Azure customer results. The key constraint is that this remains a preview: rate limits, regions, and service commitments are still unspecified.

Story brief

3 key points

xAI is adding Grok 4.6 to Microsoft Foundry’s Azure model catalog, giving enterprise teams a governed environment to compare and test it alongside competing models. The preview supports managed endpoints and enterprise controls, while Grok offers a 500,000-token context window and API pricing of $2 per million input tokens and $6 per million output tokens. The listing expands xAI’s cloud distribution after Bedrock...

  1. 01

    Foundry access is public preview, unlike Grok 4.6’s generally available Amazon Bedrock deployment announced Aug. 19.

  2. 02

    xAI positions the model for long-running agent work, with low-to-xhigh reasoning settings and a 500,000-token context window.

  3. 03

    The listed $2 input/$6 output API prices do not establish Microsoft Foundry pricing or commercial terms.

Grok 4.6 is now in Microsoft Foundry as a public preview for Azure customers. The listing gives xAI a managed route into an enterprise model catalog, where customers can test the model against alternatives in their existing cloud environment.

Foundry supplies the enterprise layer

Foundry provides managed endpoints plus enterprise security and governance controls for the Grok 4.6 preview. That gives Azure customers a platform route for evaluating and deploying the model rather than limiting access to xAI’s original release channels.

The catalog places Grok beside models from OpenAI, Anthropic, Meta, Mistral, DeepSeek, Cohere, and Hugging Face. Customers can evaluate models on their own data and select deployment options, making the listing an opportunity for workload-specific comparison rather than a claim of adoption.

A long-context model, with company-supplied positioning

Grok 4.6 has a 500,000-token context window and reasoning settings called low, medium, high, and xhigh. xAI says it is intended for sustained agent work, including navigating codebases, researching unfamiliar subjects, building applications, and checking its own output.

The commercial and evidence limits

  • xAI lists API pricing starting at $2 per million input tokens and $6 per million output tokens; those figures do not establish Foundry-specific terms.
  • Microsoft identifies benchmark results in its Foundry announcement as provider-supplied, and the announcements provide no independent Azure customer results.

Preview remains the boundary

Foundry’s public-preview status distinguishes it from the generally available Bedrock deployment announced on Aug. 19. The preview materials do not establish rate limits, regional availability, or service-level commitments. For now, the concrete result is broader access and a venue for enterprise testing; production readiness remains for customers to assess.

Editorial analysis

Our Read

Our read: xAI is treating distribution as part of the model product. Grok 4.6 moved from xAI’s own channels to GitHub Copilot and the three named cloud platforms in 14 days. Foundry changes the test from attracting teams to a standalone service to competing inside an environment where buyers can assess several providers on their own work. The next concrete signal is whether the Foundry offering moves beyond preview with disclosed operating terms, including regional coverage, rate limits, and service commitments.

Related coverage

Continue reading