Top Ten Buzzwords of 2023 Announced, AI Large Models Selected


The latest AI programming model rankings from LMArena show that Claude from Anthropic, GPT-5 from OpenAI, and Zhipu GLM-4.6 are tied for first place globally. These models, designed specifically for programming, can significantly improve the efficiency of code writing, debugging, and optimization, driving advancements in software development.
Alibaba Cloud's Tongyi series AI large models were first applied on a large scale in the e-commerce field during "Double 11", comprehensively optimizing consumer experience, business management, and traffic distribution for Taobao and Tmall. Among them, translation models such as Tongyi Qwen-MT played a core role in cross-language transactions.
The first AI large model investment competition organized by the US Nof1 institution concluded, with Alibaba's Tongyi Qianwen Qwen3-Max winning with a 22.32% return rate, demonstrating its leading capabilities in quantitative trading. The competition provided each of the six top models with $10,000 in initial funds to compete in a real trading environment on the Hyperliquid platform.
Ant Group open-sources the BaiLing Large Model Ring-flash-linear-2.0-128K, specifically targeting long-text programming. It employs a hybrid linear attention mechanism with a sparse MoE architecture, achieving performance comparable to a 40B dense model by activating only 6.1B parameters. It achieves optimal results in code generation and intelligent agent applications, efficiently addressing the challenges of long context processing.
ByteDance opens 360B-parameter Seed-OSS-36B model with 512K context window, surpassing mainstream 128K models.....