Search AI Products and News

This morning, Microsoft officially announced its latest innovative small-parameter model, Mu. This model has only 330 million parameters, yet it can match the performance of Microsoft's previously released Phi-3.5-mini, while being just one-tenth the size of Phi-3.5-mini. More notably, Mu can achieve a response speed of over 100 tokens per second on offline NPU laptops, which is a rare breakthrough in the field of small-parameter models. A major highlight of the Mu model is its support for setting intelligent features in Windows.

Microsoft has unveiled Phi-4, a new multimodal and mini-model designed to significantly improve processing capabilities across speech, vision, and text. This release represents a notable advancement in AI technology.

Kuaishou Technology Co., Ltd. announced that its independently developed visual generation large model, Kling AI, has completed a new round of capability upgrades. The upgraded Kling AI API has achieved significant improvements in virtual fitting and lip-syncing, providing stronger technical support for clients in e-commerce, advertising marketing, pan-entertainment, and AI tools.

Microsoft announced that its Phi series of generative AI models has a new member - Phi-4. Compared to previous versions, Phi-4 shows improvements in several areas, particularly in solving mathematical problems, thanks to the enhanced quality of training data. As of Thursday evening, access to Phi-4 is relatively limited, available only on Microsoft's newly launched Azure AI Foundry development platform and restricted to research purposes under the Microsoft Research Licensing Agreement. Image source note: Image generated by AI, image licensed.

Tencent has recently launched a smart workbench product called ima.copilot (shortened to ima), which is powered by Tencent's Hunyuan large model technology. This product aims to provide users with a new experience in searching, reading, and writing through its thinking knowledge base. The core features of ima.copilot include knowledge acquisition, the creation of a personalized knowledge base, and intelligent writing assistance.

Microsoft has released Phi-3.5-vision, a lightweight, multimodal open source AI model designed for processing textual and visual inputs, supporting a context length of 128K. This model is suitable for resource-constrained environments and features capabilities such as image understanding, OCR, chart parsing, and multi-image summarization, showcasing excellent performance and low latency. Comprised of 4.2 billion parameters, it is trained with high-quality data to ensure performance and privacy. It includes three models: lightweight AI, expert mix, and multimodal model, all demonstrating outstanding performance in image and video processing benchmarks.

Microsoft has released three new AI models aimed at enhancing its leadership in the multilingual and multimodal fields. The Phi-3.5 Mini Instruct is a lightweight model suitable for task execution, code generation, and logical reasoning, excelling particularly in multilingual and multi-turn conversations. The Phi-3.5 MoE (Mixture of Experts) model is rich in parameters and supports various inference tasks, especially outperforming peers in code, mathematics, and multilingual understanding. The Phi-3.5 Vision Instruct integrates text and image processing capabilities.

Microsoft has announced the launch of serverless fine-tuning capabilities for the Phi-3 small language model, enabling developers to easily adjust model performance on the Azure AI platform without managing underlying infrastructure, and it is free for initial use. The Phi-3 model, with 3 billion parameters, is suitable for enterprise developers, offering efficient performance at a lower cost. Although it has fewer parameters compared to OpenAI's GPT-3.5, it performs similarly across various applications. Serverless fine-tuning simplifies operations, reduces barriers, and supports the fine-tuning of small to medium models.

知网推出AI智能备课系统，运用AI技术和华知大模型，提供智能写作、编改、推荐和管理功能，旨在提高教学质量。此系统能让教师一键生成高质量教学方案，通过在线迭代完善教案设计，推荐与教案主题关联的学术文献，并支持教案内容的分类与导出管理，能显著提升备课效率并减轻教师工作压力。主要特点包括：增强的教学方案生成能力、高效满意的教案修改过程、精准的学术资源推荐和方便的文档管理，该系统能够对教学大纲给予智能化的支持。感兴趣的用户可以通过提供的链接进行体验和使用。

AlphaFold3（AF3）是DeepMind团队的最新增强工具，用于预测蛋白质、核酸和小分子的结构。凭借能够解析单个蛋白质序列及其复合结构的能力，AF3实现了"食谱到三维结构"的转变。其架构由三项核心模块组成：输入准备梳理序列、检索具备相似结构的分子；表示学习运用多层注意力机制更新表征；结构预测通过条件扩散模型实现预测。AF3创造性地将各类分子转化为独特数值张量，仿佛给每个分子一个独一无二的"身份证"，并运用注意力机制捕捉分子间的互相关系。在预测层面，条件扩散模型从随机噪声开始，逐步构建结构预测。训练中多种损失函数和信心评估共同作用，提高了AF3的预测准确性和可靠性，实现自我反省与提升。AF3的可视化指导和先进架构，展示了人工智能在生物学领域预测模型的革命性进步。

2024年7月19日, RWKV开源基金会全球开源了其最新最强的稠密纯RNN大语言模型RWKV-6-World14B。此模型在性能测试中体现出与拥有213亿个参数的Llama相当的英文性能，并在100多种语言与代码上表现出显著的多语言优越性，支持全球100种以上的语言和代码。RWKV-6-World14B通过基准测试显示了出众的性能，特别是在Uncheatable Eval评测中超越了Llama213B和Qwen1.514B。优点在于，该模型在训练过程中未受特殊优化困扰，基于从RWKV-4到RWKV-6的架构改进。在当前的评价排名中，该模型不仅展示了强大的实际能力，还能够在包括arXiv论文、新闻、ao3小说和GitHub代码等实时数据上进行评估，表明了其在模型能力和泛化能力上的真实性。提供下载的平台有Hugging Face、ModelScope和WiseModel等，不支持safetensor（.st）格式的Ai00用户则可以下载都在Ai00HF仓库中转换的版本。部署所需显存约为10G至28G。此款模型在多个领域展现出潜力，涉及自然语言处理、文学创作、代码处理、学术建议以及教育指令的自动化等复杂的实际应用，并且所有开源的RWKV模型在指令与对话方面有所提供，但尚无专门针对特定任务的优化训练。

小冰公司的最新发布标志着其AI数字员工产品的重大升级，新增“零样本”数字人技术（Zero-shot Xiaoice Neural Rendering，Zero-XNR）、超千亿大模型基座以及高能多媒体传输系统，显著提升实时交互质量和效果。Zero-XNR技术结合TTS语音模型，并配以高效聚类框架实现秒级高质量声音与形象复刻。大模型基座与Agent构建框架的强化集结了强大的职业交互功能，用于精准商业交互。透影音画传输系统的引入，则确保了超高清视频的传输流畅性与抗干扰性，增强了用户体验。在技术创新引领下，小冰旨在推动数字人技术的普及与产业化应用，经过微软背景的专业打磨，现已成为一家独享核心技术的中国AI公司。其技术框架涵盖了自然语言处理、语音识别、视觉交互及AI内容生成，已成功构建全球化的AI数字人产品体系，服务从金融到教育、智能汽车、智能地产等众多行业，支持数字工作者提供平稳、可靠并生产力充足的交互体验。

商汤科技与国家篮球队战略合作，通过其“日日新SenseNova5.5”大模型技术开发AI篮球产品，旨在深度整合AI与篮球、体育产业。商汤成为篮球队的人工智能战略伙伴，共同创建运动表现分析平台，利用AI技术提升运动员训练效率和竞技水平。技术通过三维模型和动作捕捉算法实时分析球员和篮球运动，生成教练决策支持，优化训练策略。在三人篮球国家队训练中应用，助力提升竞技水平及国际竞争力。AI产品不仅在竞技体育，还适用于青训体育与大众体育领域，预示智能技术成为提升运动员表现的强大驱动力。此次合作标志着智能与体育的全面深入融合，为体育强国建设贡献力量。

PaddleOCR v2.8.0发布，集前沿技术、优化项目结构、解决历史问题与增强性能于一体，树立OCR领域新标准。版本亮点包括冠军算法SVTRv2与SLANet-LCNetV2，优化算法及更新，修复关键问题，引入符合PEP518规范的教程站点。感谢社区支持，PM成员与贡献者功不可没。开发者可通过访问项目地址【https://github.com/PaddlePaddle/PaddleOCR】获取更多资源与信息。

Mistral AI was founded in May 2023. Mistral Large is comparable to GPT-4. Mistral AI launched the chat assistant Le Chat. Mistral AI's business model is similar to OpenAI. Microsoft reached a collaboration agreement with Mistral AI to continue developing in the AI field. Microsoft is developing a new small model, Phi-2. Microsoft has reached a multi-billion dollar chip deal with Intel.