## What it is
MAI-Image-2.5 is Microsoft’s new in-house image generation and editing model, shown at Build 2026 and available in Foundry. It debuted at No. 3 on Arena.ai’s image leaderboard — a +75-point jump over MAI-Image-2 — with its biggest gains in text rendering (+107) and cartoon, anime, and fantasy (+90). It’s already running inside PowerPoint Copilot.
## Why it matters
The standout is editing, not just generation. MAI-Image-2.5 handles image-to-image edits that keep faces and characters consistent across pose and style changes, full-frame restyling with object repositioning, and precise typography control — the things that break most image models. Pricing runs $47 per million image output tokens, with a cheaper Flash tier at $19.50. It’s part of Microsoft’s push to lean less on OpenAI by building its own voice, image, and transcription models in-house — a clear sign the company wants to own the full multimodal stack rather than license it.

Leave a comment