阿里发布千问3.5:性能媲美Gemini 3,Token价格仅为其1/18

新浪科技
Feb 16

  新浪科技讯 2月16日下午消息,除夕当天,阿里巴巴开源全新一代大模型千问Qwen3.5-Plus,宣布性能媲美Gemini 3 Pro,登顶全球最强开源模型。据悉,千问3.5实现了底层模型架构的全面革新,此次发布的Qwen3.5-Plus版本总参数为3970亿,激活仅170亿,以小胜大,性能超过万亿参数的Qwen3-Max模型,部署显存占用降低60%,推理效率大幅提升,最大推理吞吐量可提升至19倍。Qwen3.5-Plus的API价格每百万Token低至0.8元,仅为Gemini 3 Pro的1/18。

  与前几代的千问大语言模型不同,千问3.5实现了从纯文本模型到原生多模态模型的代际跃迁。千问3预训练在纯文本Tokens上进行,而千问3.5则基于视觉和文本混合token上预训练,并大幅新增中英文、多语言、STEM和推理等数据,让张开“眼睛”的大模型学会了更密集的世界知识和推理逻辑,以不到40%的参数量获得超万亿的Qwen3-Max基座模型的顶尖性能,在推理、编程、Agent智能体等全方位基准评估中均表现优异。

  据悉,千问3.5在MMLU-Pro知识推理评测中得分87.8分,超越GPT-5.2;在博士级难题GPQA测评中斩获88.4分,高于Claude 4.5;在指令遵循IFBench以76.5分刷新所有模型纪录;而在通用Agent评测BFCL-V4、搜索Agent评测Browsecomp等基准中,千问3.5表现均超越Gemini 3 Pro、GPT-5.2。(文猛)

海量资讯、精准解读,尽在新浪财经APP

责任编辑:尉旖涵

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10