阿里云近日宣布，其通义千问多模态大模型Qwen-VL再次升级，推出了Max版本。这一升级版模型在视觉推理能力和中文理解能力方面得到了显著提升，能够根据图片进行识人、答题、创作以及写代码等任务。据悉，Qwen-VL-Max的整体性能已经堪比GPT-4V和谷歌Gemini Ultra,甚至在多个权威测评中取得了佳绩。
Qwen-VL-Plus和Qwen-VL-Max在MMMU、MathVista等测评中远超业界所有开源模型，在文档分析(DocVQA)、中文图像相关(MM-Bench-CN)等任务上更是超越了GPT-4V,达到了世界最佳水平。这一系列的成果表明，阿里云在人工智能领域的研究实力和技术创新能力已经达到了世界领先水平。
阿里云表示，未来将继续加大对多模态大模型的研究力度，不断提升模型的性能和应用场景，为用户提供更加智能化、便捷化的服务。同时，阿里云也希望通过与各界合作，共同推动人工智能技术的发展和应用，为社会带来更多的价值和便利。
英语如下：
Title: Aliyun’s Qwen-VL Upgraded, Performance Surpasses GPT-4V and Google Gemini
Keywords: Aliyun, Tongyi Qianwen, Large Model Upgrade
Aliyun recently announced that its Tongyi Qianwen multimodal large model, Qwen-VL, has been upgraded again, introducing the Max version. This upgraded model has seen significant improvements in visual reasoning and Chinese comprehension capabilities, allowing it to perform tasks such as recognizing people based on images, answering questions, creating content, and writing code. It is reported that the overall performance of Qwen-VL-Max has surpassed that of GPT-4V and Google Gemini Ultra, achieving outstanding results in various authoritative evaluations.
Qwen-VL-Plus and Qwen-VL-Max far exceeded all open-source models in industry benchmarks such as MMMU and MathVista. In tasks like document analysis (DocVQA) and Chinese image relatedness (MM-Bench-CN), they even surpassed GPT-4V, reaching the world’s best level. These achievements demonstrate Aliyun’s leading research and technological innovation capabilities in the field of artificial intelligence.
Aliyun stated that it will continue to increase its research efforts on large multimodal models, continuously improving their performance and application scenarios to provide users with more intelligent and convenient services. At the same time, Aliyun hopes to collaborate with various sectors to jointly promote the development and application of artificial intelligence technology, bringing more value and convenience to society.
【来源】https://news.mydrivers.com/1/960/960575.htm
Views: 1