商汤发布 SenseNova U1.5-Lite-Preview 开源模型
商汤推出 SenseNova U1.5-Lite-Preview，一个基于 NEO-Unify 架构的轻量级原生统一多模态模型，仅 8B-MoT 参数即可达到商业闭源模型的生成与编辑质量。
Introducing SenseNova U1.5-Lite-Preview.
An early open-source preview of our lightweight， natively unified multimodal model – built on the native NEO-Unify architecture to natively understand， reason， generate， and edit across modalities. With just 8B-MoT parameters， it delivers generation and editing quality that rivals commercial closed-source models.
🌟This preview brings：
🔹Up to 4K resolution， delivering richer details and fewer visual artifacts
🔹Stronger EN/CN text rendering and complex layout composition
🔹More precise visual control with long， multi-constraint prompts
🔹More reliable native image editing across reference-guided style transfer， multi-reference composition， local infographic editing and text editing
🌟Improvements over U1 on generation & editing benchmarks：
🔹Qwen-Image-Bench： 47.14 → 55.20 （w/ PE）
🔹ImgEdit-Bench： 3.90 → 4.37
🔹GEdit-Bench-en： 7.47 → 8.17
🔹GEdit-Bench-zh： 7.42 → 8.05
From U1 to U1.5， this open-source journey reflects our continued innovation at the foundation-model level. Meanwhile， U1 Pro， our production-ready model for creators， will open for public access soon.
🛠️ GitHub： https://github.com/OpenSenseNova/SenseNova-U1/blob/feat/u1.5_preview/docs/u1.5_preview.md
🤗 Hugging Face： https://huggingface.co/sensenova/SenseNova-U1.5-8B-MoT-Preview
👾 Discord： https://discord.com/invite/BuTXPHmQub