deepseek-v4.1POST

DeepSeek V4.1-Flash-1017 launch: scores above Opus 5 and GPT 5.6-Sol

Launch summary: DeepSWE 77.0, Apex Bench 72.1, Agents' Last Exam 35.4, claimed to beat Opus 5 and GPT 5.6-Sol, with DeepSeek Harness and CodeBuddy support.

mio

@miomm0766Archived 11 days agoOriginally posted 2026.10.17

mio @miomm0766Originally posted 2026.10.17

Open original on X

⚡ 刚刚DeepSeek-V4.1-Flash-1017 正式上线！核心亮点一句话总结：性能更强、速度更快、输出更短，完美实现性能与成本的帕累托改进。智能体与编程表现亮眼：在 DeepSWE (77.0)、Apex Bench (72.1)、Agents' Last Exam (35.4) 等多项测试中正面对决并超越 Claude Opus 5 与 GPT 5.6-Sol。工具生态深度适配：专为 @DeepSeekHarness、腾讯 CodeBuddy、WorkBuddy 等优化。即刻可用：目前已全面登陆官网、App 及 API。附上官方最新跑分对比图 👇 #DeepSeek #AI新闻 #大模型

This post quotes another post: x.com/deepseek_ai/status/2111375640717033777

x.com/miomm0766/status/2111475822993510626Archived 2026.10.18 16:36