Microsoft is replacing outside AI models in its own products with cheaper in-house alternatives
Microsoft is replacing outside AI models in its own products with cheaper in-house alternatives
At Ignite, Microsoft AI unveiled seven new in-house models spanning code, security, images, speech, routing and retrieval, positioned on price-to-performance against OpenAI's lineup. Microsoft says MAI Code 2 averages 98 percent of GPT-6.1 Sol's performance across four benchmarks at 40 percent of the cost, and MAI Cyber 2 Flash solves 92.3 percent of CyberGym-E2E tasks versus 91.5 percent for GPT-6 Astra at under 2 percent of the per-task cost. The company also introduced MAI Licensed Weights, letting enterprises fine-tune MAI models on their own data in their own Azure environment, with launch date and pricing still pending.
The article details Microsoft's pricing and benchmark claims for a broad in-house model lineup, offering a concrete view of how it is positioning itself against OpenAI and other providers.
Source: The Decoder · the-decoder.com