1、谷歌发布Gemma-2-2B
仅2B参数,可在手机端流畅运行,在LLM竞技场超过了GPT-3.5和Mixtral-8x7b
模型地址:https://huggingface.co/collections/google/gemma-2-2b-release-66a20f3796a2ff2a7c76f98f
2、手办风格AI绘画模型
模型地址:https://civitai.com/models/400329/pvc-style-modelmovable-figure-model-pony

3、阿里CosyVoice更新
TTS模型,更新支持性别、年龄、个性设定;新增支持生成语气音,例如笑声、咳嗽、呼吸;
Github:https://github.com/FunAudioLLM/CosyVoice