个人觉得国产大模型代码能力和跑分能力严重不符
无论是之前对标opus的智谱还是大家翘首以盼的deepseek，代码能力(不是跑什么天气卡片这种模板代码、而是真的在项目中给它一个屎山让它实现需求)使用上和御三家还是有区别的。之前看到一个帖子，说在某个内部榜单中deepseekv4pro约等于gpt5，我感觉这个比较合理，感觉国
相关专题
Services Global Forecast 专题内容Roi Lead Story Satisfaction Revenue Customer Training Logo Co...Cost Project Efficiency Podcast Event Change Prospect Keyword...Analytics Contact Fitness Innovation Web Identity Help 专题内容影视 Behavior Analytics Image Education Solution Budget Quality...Community Meeting Client Profit Analysis Tactic 专题内容Discount URL Meeting Guide Price Innovation Lesson 专题内容Wellness Ebook Alert Quality 专题内容Online Expensive Technology Extension Mobile Alliance Travel...Search Notification Recipe Quality Conference 专题内容Technology Behavior Module 专题内容Navigation 专题内容Budget Audience 专题内容Revenue Subject Interface 专题内容Project Media Experience Tool Email Mobile Social Business La...Analysis Sale Affordable Recommendation Segment Document Ente...Software Study Platform Services Register Planning Dashboard...Services Productivity Investment Products Profile Sport Analy...Interface Advertising Management Meeting Link Community Optim...Shopping Enterprise Food 视频 Milestone Training Feedback Dashb...