deepseek-v4.1POST
V4.1-Flash-1017 开源权重：DeepSWE 77.0 超 Opus 5 与 GPT 5.6-Sol
帖称该版本开源权重、回答更短，厂商自报 DeepSWE v1.1 得分 77.0，高于 Claude Opus 5 的 74.0 和 GPT 5.6-Sol 的 73.0，但作者提醒需自测再切换。
pcherkashin.x
@pcherkashinX12 天前收录原推发布于 2026.10.18
pcherkashin.x @pcherkashinX原推发布于 2026.10.18
@lydiahallie DeepSeek V4.1-Flash-1017 is out with open weights and shorter answers. On DeepSWE v1.1 it scores 77.0, against 74.0 for Claude Opus 5 and 73.0 for GPT 5.6-Sol. Those are vendor numbers. I'd run it on my own tasks before switching. https://t.me/ai_machinelearning_big_data/11147
x.com/pcherkashinx/status/2111717677924303148采集于 2026.10.18 16:34