Microsoft Launches MAI-Thinking-1 Reasoning Model in Azure Foundry Public Preview
Microsoft has released MAI-Thinking-1, its first dedicated reasoning model, into public preview via Microsoft Azure Foundry. Unlike standard chat models, it was built from scratch using a Mixture-of-Experts architecture designed for multi-step planning, tool use, and extended context reasoning. The model is competitively priced at $2 per million input tokens and $8 per million output tokens, making it viable for production agent workloads. It benchmarks well on SWE-Bench Pro against higher-cost models in its class, and Microsoft confirms it was trained on clean data without distillation from third-party models. Developers using C# can integrate it through the Microsoft.Extensions.AI IChatClient interface, requiring only a configuration change to swap it in alongside existing Foundry models like GPT-4o.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in