MAI-Image-2.5-Pro
MAI-Image-2.5-Pro
Vintage: 2026-08. Primary evidence is an independent @ArtificialAnlys X post dated 2026-08-19. Not a Microsoft blog. Animated gif on the post was not transcribed (not video).
One-line summary: Microsoft AI's quality-focused image model; per Artificial Analysis on X, it debuts #1 on the Image Editing Leaderboard and #7 on Text to Image, with Foundry per-token prices as written in that post.
What it is
A Microsoft AI image model in a three-point family (MAI-Image-2.5-Pro, MAI-Image-2.5, MAI-Image-2.5-Flash) that the bench-org post says Microsoft positions across a quality-speed-cost curve. The post says Pro launched July 23 in preview on Microsoft Foundry and is aimed at hero imagery, detailed editing, and precise in-image text rendering. This page records the Artificial Analysis X post only.
Why it matters to this thread
Frontier / first-party model releases and independent benches are in-scope. This is the first dated MAI-Image-2.5-Pro print the thread has. Treat it as a bench-org X post, not a Microsoft announcement. Image generation/editing is also the domain of threads/ai-video-generation; this ingest stays in the AI thread because the user routed the weekday X pass here. Do not invent a Microsoft blog.
Key facts (from 2026-08-20-x-ai-news-20-aug-2026-openai-private-safety-claude-connectors)
All bullets are one X post (fetch_method: x-mcp) by @ArtificialAnlys, 2026-08-19. Permalinks live on the source page.
- Ranks (as written): "Microsoft's MAI-Image-2.5-Pro debuts at #1 on the Artificial Analysis Image Editing Leaderboard, and takes the #7 spot in Text to Image." Same post: in the Image Arena it "debuts at #1 on the Image Editing Leaderboard, surpassing Reve 2.1, GPT Image 2, and Microsoft's own MAI-Image-2.5."
- Positioning (as written by the bench org, attributing Microsoft): quality-focused; "highest-fidelity image model to date"; family covers quality-speed-cost; launched July 23 in preview on Microsoft Foundry; available in preview across seven global-standard regions; tryable in the MAI Playground.
- Foundry prices (as written): "$5 per 1M text input tokens, $8 per 1M image input tokens, and $106 per 1M image output tokens, which works out to roughly $108.5 per 1k 1024x1024 images." Comparables in the same post: "$48 per 1k images for MAI-Image-2.5 and $20 per 1k for MAI-Image-2.5-Flash."
- Qualify: independent bench-org X post, not a Microsoft blog.
has_media: animated_gif— not video; do not transcribe.
What this source does not establish
- Not a Microsoft blog or model card. Do not treat @ArtificialAnlys as microsoft speaking.
- No sibling entity pages for MAI-Image-2.5 or MAI-Image-2.5-Flash — they are named in one post as family points.
- No Reve / GPT Image 2 entity pages — named only as models Pro surpassed on the editing board.
- Gif was not transcribed.
Open questions
- Do the #1 editing / #7 T2I ranks hold on a later Arena snapshot, or only at debut?
- What does an official Microsoft write-up add (eval tables, safety, exact region list) that this X post omits?