2025년 12월 5일 | Tongyi Weekly: Tongyi Lab에서 제공하는 최첨단 AI를 매주 만나보세요

Source: Dev.to

Hello, builders and visionaries,

This week, local AI got a major upgrade — and your workflows just got sharper, faster, and more expressive. Let’s dive in.

Ecosystem Highlights

Qwen3-Next Lands on llama.cpp

Big news for local AI enthusiasts: llama.cpp (PR #16095) just added support for Qwen3-Next — Qwen’s new hybrid architecture! You can now run Qwen3-Next locally with efficient CPU/GPU inference.

🔗 View the PR

Z-Image Milestone: 1 Million ComfyUI Workflow Downloads in One Week

Our Z-Image ComfyUI workflow just crossed 1,000,000 downloads in 7 days — a historic moment for open generative AI. Thank you for making Z-Image not just fast and small — but beloved.

🔗 Learn more

Z-Image‑Turbo ControlNet Union Now Live

Alibaba‑PAI just dropped a Z-Image‑Turbo ControlNet Union that handles Canny, Depth, Pose, HED, and MLSD in a single model. Trained on 1 million images at 1328 px resolution, so it actually respects high‑res details.

🔗 Download on Hugging Face

Community Spotlights

Concept Sliders for Z-Image: sliders-for-windows from sdbds (Qing Long)

Want more texture without changing your subject? sdbds drops this LoRA for Z-Image using Concept Sliders to dig deeper into the model. It enhances fine details and lighting rendering while keeping your original content 100 % intact—no hallucinations, just high‑definition polish.

👉 Try it here

Children’s Drawings: z_image_turbo_childrens_drawings from ostris

ostris released this “Children’s Drawings” model as part of a full training tutorial with AI Toolkit. Perfect for when you need that “my 5‑year‑old drew this” energy in your generations.

👉 Try it here

Technically Color Z: Technically-Color-Z-Image-Turbo from renderartist

Technically Color Z by renderartist captures the unmistakable essence of classic film. This LoRA greatly enhances depth and brilliance of hues, creating realistic yet dreamlike textures, lush greens, brilliant blues, and the distinctive glow seen in classic productions.

👉 Try it here

Pixel Art: elusarca-pixel-art-style-lora-zimage-turbo from reverentelusarca

If you’re generating retro assets, you need this in your workflow. The Pixel Art LoRA for Z‑Image‑Turbo cleans up noise and forces a crispier grid than the default output—a must‑have for that 16‑bit aesthetic.

👉 Try it here

Material Transfer: Qwen-Edit-2509-Material-transfer from oumoumad

oumoumad dropped a Material Transfer LoRA for Qwen‑Image‑Edit‑2509. Feed a plain render (e.g., a car seat or cabin) and a material board or shaderball, and it applies the texture instantly. CMF workflow just got a whole lot faster.

👉 Try it here

Want More? Stay Updated

Every week, we bring you:

• New model releases & upgrades
• AI research breakthroughs
• Open‑source tools you can use today
• Community highlights that inspire

👉 Subscribe to The Tongyi Weekly

Thank you for being part of this journey.

Tongyi Lab is a research institution under Alibaba Group dedicated to artificial intelligence and foundation models, focusing on the research, development, and innovative applications of AI models across diverse domains. Its research spans large language models (LLMs), multimodal understanding and generation, visual AIGC, speech technologies, and more.