Microsoft unveils MAI-Image-2.5-Pro and MAI-Voice-2-Flash, slashes AI costs

Microsoft announced two new in‑house AI model variants, MAI‑Image‑2.5‑Pro and MAI‑Voice‑2‑Flash, in public preview. MAI‑Image‑2.5‑Pro is a high‑fidelity image generator priced at $5 per million text tokens, $8 per million image inputs and $106 per million image outputs, while MAI‑Voice‑2‑Flash delivers speech at $15 per million characters, runs twice as fast as its predecessor and costs 32 % less. Both models are already deployed in Microsoft 365 products—PowerPoint, Bing Image Creator, OneDrive, Excel, GitHub Copilot, Dynamics 365 and Azure—replacing OpenAI and Anthropic models. Microsoft reports GPU cost reductions of 84 % for image generation in PowerPoint and up to 89 % for voice workloads, along with tighter control over latency, moderation and compliance. The rollout is part of a broader strategy to reduce reliance on external providers, improve margins, and support an open‑weight AI ecosystem, underscored by an open letter signed with other tech firms.