Now in Foundry: Tongyi-MAI Z-Image-Turbo, with FLUX.1-schnell and SDXL base 1.0

2026-05-18 · Source: Microsoft Foundry Blog articles · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

Microsoft Foundry has integrated three new image generation models from the Hugging Face collection as of May 18, 2026. These include Tongyi-MAI's Z-Image-Turbo, a 6B parameter model designed for low-latency, 8-step inference on a single GPU with native bilingual text rendering (English and Chinese) and up to 1024x1024 resolution. Also added is Black Forest Labs' FLUX.1-schnell, a 12B rectified flow transformer known for 1-4 step inference, flexible resolution up to 2 megapixels, and permissive Apache 2.0 licensing. Finally, Stability AI's stable-diffusion-xl-base-1.0 (SDXL), a 2.6B UNet model (approx. 3.5B total) with a dual text encoder design and CreativeML Open RAIL++-M licensing, is now available. These models can be deployed via the Foundry model catalog or directly from the Hugging Face Hub.

Key takeaway

For Computer Vision Engineers developing image generation applications, you should evaluate these newly integrated models in Microsoft Foundry. Z-Image-Turbo offers unique bilingual text rendering and low-latency inference, while FLUX.1-schnell provides rapid generation under a permissive license. SDXL base 1.0, with its dual text encoders, remains a robust option for high-quality image synthesis, offering flexibility for various commercial projects.

Key insights

Microsoft Foundry now offers three advanced image generation models for diverse commercial and technical applications.

Principles

• Single-stream architectures can enhance parameter efficiency.
• Rectified flow transformers enable rapid, low-step inference.
• Dual text encoders improve semantic and token-level alignment.

Method

Z-Image-Turbo uses Scalable Single-Stream Diffusion Transformer (S3-DiT) architecture and Decoupled Distribution Matching Distillation (Decoupled-DMD) with DMDR for efficient 8-step inference.

In practice

• Use Z-Image-Turbo for bilingual text rendering in marketing assets.
• Deploy FLUX.1-schnell for latency-sensitive image generation.
• Combine SDXL base with its refiner for enhanced image quality.

Topics

• Microsoft Foundry
• Hugging Face Models
• Tongyi-MAI Z-Image-Turbo
• FLUX.1-schnell
• Stable Diffusion XL

Code references

• Stability-AI/generative-models

Best for: Computer Vision Engineer, AI Engineer, Machine Learning Engineer, AI Product Manager

Related on AIssential

• FINALLY! NEW FREE NSFW IMAGE AI KING! LESS THAN 8GB VRAM! — Z Image Base offers enhanced versatility and fine-tuning for text-to-image generation, even on limited VRAM.
• Meet FLUX.2 [flex] for Text‑Heavy Design and UI Prototyping Now Available on Microsoft Foundry — FLUX.2 [flex] is a specialized generative AI model for precise text and graphic design within creative workflows.
• Best Free Image Generators on Hugging Face Right Now! — Hugging Face hosts diverse, free, high-quality image generation models, each suited for distinct commercial, artistic, or technical use cases.
• ComfyUI Update: What’s new from the last few weeks, SD Turbo, Stable Zero123, Group Nodes, FP8, and more. — ComfyUI updates enhance performance, memory efficiency, and creative capabilities through new models and UI features.
• TrioPose: Native Triple-Stream Diffusion Transformers for Pose-Guided Text-to-Image Generation — TrioPose natively integrates pose as an independent modality within diffusion transformers to resolve multi-person image generation distortions.
• RefGC-SR$^2$: Reference-guided Generated Content Super-Resolution and Refinement — Reusing high-resolution references post-generation can simultaneously recover details, refine artifacts, and upscale generated content.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.