WHAT NEW IN X10.31 SAT

AI collections from X · 22 collections · 63,170 posts

deepseek-v4.1POST

DeepSeek V4.1-Flash-1017 cost report: DeepSWE 77 at about $0.20 per task

The post says V4.1-Flash-1017 scores 77 on DeepSWE at roughly $0.20 per task (about $0.10 in low settings), weights are on Hugging Face, and DeepSeek claims it beats Opus 5 and GPT 5.6-Sol pending independent tests.

ContextBaron
@ContextBaronArchived 12 days agoOriginally posted 2026.10.19
ContextBaron @ContextBaronOriginally posted 2026.10.19
Open original on X

Grok Bot 這次更新,最實際的是電腦任務平均快 3 倍,其他回覆時間少於一半。速度令你肯把更多事交出去。 同一時間 DeepSeek 又把 agentic coding 單價壓低。V4.1-Flash-1017 自家 DeepSWE 77,約 $0.20/task,低設定約 $0.10。權重已上 Hugging Face,公司自報勝過 Opus 5 和 GPT 5.6-Sol,獨立評測未出。 一邊是 agent 產品變快變日用,一邊是開源 Flash 繼續壓任務成本。你選的是整條 workflow,不是單一模型分數。

This post quotes another post: x.com/bot/status/2111184275374813362

x.com/contextbaron/status/2112017288497103257Archived 2026.10.19 12:03