Microsoft and xAI have added Grok 3, xAI’s flagship large language model, to Azure AI Foundry and GitHub Models. Starting May 19, developers can try it free for two weeks on both platforms, as part of a push to broaden access to high-performance AI models in enterprise settings.
On Azure, Grok 3 joins a diverse catalogue that already includes offerings from OpenAI, Meta, Cohere, and Hugging Face. With xAI’s Colossus-trained model now in the mix, Microsoft is reinforcing its vision of a multi-model ecosystem. Meanwhile, GitHub Models has added Grok 3 to its public preview, giving users an easy way to experiment through its playground or integrate it via API.
Built for reasoning, coding, and visual understanding, Grok 3 comes with an extended 131K token context window and support for structured outputs and external tools. These features suit it for complex tasks in healthcare, finance, law, and science. Benchmarks shared by xAI show it performing competitively across areas like code generation, instruction following, and agentic workflows.
Grok 3 has been trained on Colossus, which xAI claims is the world’s most powerful AI training system. Elon Musk earlier said that the cluster was built in 122 days from start to finish.
Azure AI Foundry offers two deployment modes: a serverless pay-as-you-go model and a Provisioned Throughput (PTU) mode for predictable, high-volume use. The official blog post stated, “Grok 3 will soon support PTU deployments on Azure. This means you can reserve a fixed throughput capacity for the model, allowing you access to dedicated resources to handle a certain rate of requests.”
Pricing kicks in from early June, starting at $3 per million input tokens and $15 for output. In the meantime, the free preview offers a risk-free test window.
While xAI describes Grok 3 as “designed to accelerate scientific discovery”, its debut across Microsoft platforms sounds like an interesting integration. Whether one is building AI agents or parsing lengthy legal contracts, Grok 3 now sits within arm’s reach—on Azure for scale, or GitHub for quick-start prototyping and fixing.
