微软推出了AI图像模型MAI-Image-2,现在可以预览,并改进现实性和文本生成. Microsoft launches MAI-Image-2, its new AI image model, now available in preview with improved realism and text generation.

微软推出了其第二代AI图像模型MAI-Image-2,现在可以通过MAI Playground进行预览,并将向Copilot和Bing Image Creator推出. Microsoft has launched MAI-Image-2, its second-generation AI image model, now available in preview via the MAI Playground and rolling out to Copilot and Bing Image Creator. 模型在Arena.ai排名榜上位列第三,展示了改进的摄影现实主义,精确的皮肤色调,自然照明和更好的图像中文本生成. The model ranks third on the Arena.ai leaderboard, showcasing improved photorealism, accurate skin tones, natural lighting, and better in-image text generation. 设计的目的是减少后期制作工作,并支持复杂场景创作. Designed with input from creatives, it aims to reduce post-production work and supports complex scene creation. 通过企业用户的API扩展访问,并将很快通过微软 Foundry提供. Access is expanding through API for enterprise users and will soon be available via Microsoft Foundry. 这一发布标志着微软不再依赖OpenAI来生成图像, 并反映了由内部人工智能研究和新的GB200计算基础设施推动的更快创新. The release marks Microsoft’s shift from relying on OpenAI for image generation and reflects faster innovation driven by in-house AI research and new GB200 compute infrastructure.