AI模型精选

DeepSeek-V4-Flash 五项信号公布:总体+1.98%,确认成功+6.99%

DeepSeek-V4-Flash-20260731(High) across 5 key signals: - Overall: +1.98% - Confirmed Success: +6.99%...

精选理由

想知道 DeepSeek 最新版行不行?看这五个真实用户信号,总体涨近2%,确认成功涨7%,但可引导性跌了。

AI 摘要

LMArena 公布 DeepSeek-V4-Flash-20260731(High) 的五项关键信号。总体得分上升 1.98%,确认成功项提升 6.99%,好评对差评比增加 1.54%。可引导性下降 2.31%,Bash 恢复得分 2.53%,工具幻觉率 1.2%。数据来自竞技场真实用户反馈。

原文 · lmarena.ai

DeepSeek-V4-Flash-20260731(High) across 5 key signals: - Overall: +1.98% - Confirmed Success: +6.99%...

DeepSeek-V4-Flash-20260731(High) across 5 key signals: - Overall: +1.98% - Confirmed Success: +6.99% - Praise vs Complaint: +1.54% - Steerability: -2.31% - Bash Recovery: 2.53% - Tool Hallucination: 1.2% 💬 1 🔄 0 ❤️ 14 👀 1175 📊 2 ⚡