Microsoft swaps external models for cheaper in-house AI across its products
Microsoft is replacing outside AI models in its own products with cheaper in-house alternatives
At Ignite, Microsoft released seven new models covering coding, security, image, speech and routing, and is replacing external models with its own across its products. Microsoft says MAI Code 2 averages 98% of GPT-6.1 Sol's performance on four benchmarks at 40% of the cost, a comparison that includes a 50% promotional discount on its own model running through May 17, 2027.
Why it matters: Microsoft compares its models with external ones on cost per task, and the code model's price edge depends on a promotional discount with a set end date.