Skip to content
极客公园·· 2026-10-20AI score22

DeepSeek-V4.1-Flash-1017 tops Code Arena WebDev open-source model rankings; Doubao debunks 55% compute claim for micro-dramas; WeChat launches incentive for adapting official-account novels into micro-dramas | Geek Morning Brief

DeepSeek 新模型登顶开源模型榜首;国内 55% 算力用于视频和短剧?豆包发言人辟谣;微信发起公众号小说改编微短剧激励|极客早知道

AI briefing

DeepSeek-V4.1-Flash-1017 (Max) scored 1670, surpassing Kimi K3 (Max) to become the highest-ranked open-source model on the Code Arena WebDev leaderboard, just 12 points behind GPT-6 Sol (Max).

Open original geekpark.net