deepseek-v4.1IMAGE
DeepSeek V4.1-Flash-1017: 11 benchmarks up, DeepSWE 77 beats Opus 5's 74
A roundup says V4.1-Flash-1017 improved on all 11 benchmarks, with DeepSWE 77.0 above Claude Opus 5's 74.0 and Terminal-Bench up from 31.2 to 42.3, plus MIT-licensed weights and a Search & Fetch API.
logang
@logang2008Archived 12 days agoOriginally posted 2026.10.18
logang @logang2008Originally posted 2026.10.18
DeepSeek V4.1-Flash-1017 上线：11 项基准全涨。DeepSWE 77.0 超 Claude Opus 5 的 74.0，Terminal-Bench 从 31.2 涨到 42.3，输出更短、跑得更快。权重 MIT 开源，还开放了 Search & Fetch API。开源模型正用性价比挤压闭源溢价。https://www.techi.com/deepseek-v4-1-flash-1017-update-benchmarks-price/ #DeepSeek #开源模型 #AI
x.com/logang2008/status/2111698517320171902Archived 2026.10.18 16:34