Microsoft MAI-Image-2.6-Preview 获图像编辑基准第三
微软 MAI-Image-2.6-Preview 图像编辑排第三,文本渲染和商业设计分数大涨。
Microsoft 发布的 MAI-Image-2.6-Preview 在 Arena.
ai 单图编辑基准中获得 1420 分排名第三。该模型超越了 Google 的 Nano Banana 和 Meta 的 Muse Image,比前代 MAI-Image-2.
5 提升了 19 分。目前该模型落后于排名第二的 Grok Imagine Image 2.
0 19 分,落后于排名第一的 GPT Image 2 43 分。MAI-Image-2.
6-Preview 在文本渲染和商业设计类别分别提升了 43 分和 38 分。
We hit No.3 on the Image-editing benchmark! ... beating Google’s Nano Banana and Meta’s Muse Image models. Arena.ai @arena MAI-Image-2.6-Preview by @MicrosoftAI has landed at #3 in Single Image Edit with 1,420 pts! This is +19 pts above MAI-Image-2.5, moving from #5 to #3 . It sits 19 pts behind preliminary Grok Imagine Image 2.0 (low) at #2 and 43 pts behind GPT Image 2 (medium) at #1 . By category, MAI-Image-2.6-Preview is #3 in across: - Product, Branding & Commercial Design - 3D Imaging & Modeling - Cartoon, Anime & Fantasy - Photorealistic & Cinematic Imagery #4 in Portraits and #8 in Text Rendering. The biggest jumps over MAI-Image-2.5 are in Text Rendering (+43 pts) and Product, Branding & Commercial Design (+38 pts). Congrats again to the @MicrosoftAI team on this release! 🔗 View Quoted Tweet📊 10 ⚡
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include MicrosoftAI, MAI-Image-2.6-Preview, Google, Meta.
Source attribution: curated from Mustafa Suleyman (@mustafasuleyman). Cross-source corroboration: 2 independent sources have covered this development (including lmarena.ai), confirming industry-wide signal strength.
- lmarena.ai08-18 17:16Source