Qwen Image 2.1 晨报 2026-10-29
本期审阅 2 个 Qwen 2.1 新版本，没有 S / A 条目。Flat Cutie Style 是扁平可爱动漫画风的编辑类 LoRA 移植，有完整触发句但说明只有一行，定为 B；“Lighting Fast Head Swap”只是一个 ComfyUI 工作流 JSON，模型卡主要是 RunningHub 邀请…
ZZAI: AI Canvas Agent, video and image creation platform
ZZAI (www.zzai.app) brings Canvas Agent, AI video generation, AI image generation, motion video, a prompt gallery and AI Radar news into one workspace. Describe an idea and create with Seedance, GPT Image, Kling, Vidu and other leading models; developers reach the same models through an OpenAI-compatible API. Pay as you go, no membership.
AI models and pricing · Prompt gallery · AI Radar · API docs
One platform for the whole creative workflow
Canvas Agent
Describe what you need and the agent lays out prompts, references, image and video nodes on an infinite canvas. Every step stays editable.
AI video generation
Text to video, image to video and multimodal references with Seedance 2.5, Kling, Vidu, Wan, MiniMax H3 and more.
AI image generation
Create posters, product shots, portraits and illustrations with GPT Image 2.5, Seedream 5 Pro and more.
Motion video
Turn words, a logo or numbers into intros, title cards and animated data videos.
Prompt gallery
Browse real generations, copy the full prompt and settings, and reuse them on the canvas in one click.
AI Radar
Daily news on AI models, tools, papers and open-source projects, synced to this page automatically.
Developer API
One OpenAI-compatible API for asynchronous video tasks, images, audio and language models.
AI Radar: latest updates
MOSS Image 晨报 2026-10-29｜首推 Retro Anime Style
本期审阅 45 个 MOSS 新版本，定为 A 5 个、B 11 个、C 29 个，无排除。首推 Retro Anime Style：八九十年代 OVA 式复古动画画风，有触发词和权重区间、示例多，因缺采样参数由 S 降为 A。其他亮点：可正可负的年龄滑杆、怪诞绘本风 Cartoon World、参数写全的写实自拍感…
MiniMax H3 晨报 2026-10-29｜首推 MiniMaxH3_Enhancers
本期审阅 10 个 H3 新版本。首推 MiniMaxH3_Enhancers 的 Epoch10：无需触发词的清晰度加美颜增强 LoRA，两遍生成流程参数和前后对照示例都给全，因美颜会改动人脸由 S 降为 A；同模型 Epoch5 是更轻的一档，定为 B。另一亮点是 Seamless Anime VAE，替换官方…
Flux.3 Klein 晨报 2026-10-29
本期审阅 2 个 Flux.3 Klein 新版本，没有 S / A 条目。Woodland Camo BDU 是美军林地迷彩作训服服装 LoRA，作者给了强度下限和多人分别写名牌的提示词写法，定为 B；Prompt FrameWork 只是一个 6 KB 的提示词压缩包、与 Flux.3 Klein 本身无关，由…
MOSS Image Evening Report 2026-10-28 | Top Pick: MOSS Pixel Art
This issue reviewed 44 new MOSS versions, rated as S 1, A 4, B 14, C 24, with 1 excluded. Top pick: MOSS Pixel Art v0.3: A LoRA for pixel art / 2D game assets…
MiniMax H3 Evening Report 2026-10-28 | Top Pick: KinetiX · Ground Fracture & Impact Dynamics Slider
This issue reviews 2 new H3 versions. Top pick: KinetiX Ground Fracture & Impact Feedback Slider: uses weight control to manage impact craters, cracks, and deb…
Featured gallery cases
1990s Cool-Tone Xianxia Portrait in Emerald-Green Hanfu · gpt-image-2.5
Ultra-Detailed 3D Comic Star Collectible Poster · gpt-image-2.5
Retro Racing Poster Design Prompts · gpt-image-2.5
Chibi Character Chat Sticker Generation Prompt · gpt-image-2.5
Children's Science Encyclopedia Poster: Discovering Bees · gpt-image-2.5
Fashion Color Block Collage Portrait Photography · gpt-image-2.5
Portrait of a Woman in a Purple Sari by a Red Brick Wall in Golden Sunlight · gpt-image-2.5
Hyperrealistic Luxury Fashion Portrait · gpt-image-2.5
Frequently asked questions
- What is ZZAI?
- ZZAI (zzai.app, also known as ZhenZhen AI) is an AI creation platform for creators and developers, offering Canvas Agent, AI video and image generation, motion video, a prompt gallery, AI Radar news and a unified model API.
- How does ZZAI pricing work?
- You pay only for what you use, with no membership. Every model's price is listed on the AI models and pricing page.
- Which AI models are supported?
- Video, image, audio and language models including Seedance, GPT Image, Seedream, Kling, Vidu, Wan, MiniMax and DeepSeek, with new models added regularly.