“This is why NVIDIA’s QAD quants matter for local AI. ”
Loading the current collection. Item counts are calculated when read, so they reflect the archive now.
AI collections from X · 22 collections · 63,170 posts
DeepSeek-V4.1-Flash-1017, released October 17, 2026 with open weights: official specs and API pricing, hands-on tests, same-task comparisons with Opus 5.5, Gemini 4 Argon and GPT-6, coding and agent use, running the open weights locally, real cost data, and complaints from X.
“This is why NVIDIA’s QAD quants matter for local AI. ”
“DeepSeek-V4.1-Flash-20261017 (Max) just hit #1 among open models…”
“DeepSeek called it a "minor update". Terminal-Bench went from 31…”
“DeepSeek又发了个新模型,V4.1-Flash。”
“DeepSeekのAPI、平日の日本時間10〜13時と15〜19時はピーク料金で、夜や土日の2倍になる。”
“DeepSeek明年如果真能打到Opus 5.5水平,AI 投资界天要塌了 ”
“http://linux.do 10-19 摘要 ”
“Se o upgrade do DeepSeek V4.1 flash é uma destilação do Astra, v…”