2026-07-21 digest

23 items on this page: 3 with their own page, 20 short.

A digest puts the short items from one day or one week on a single page. Opened alone, those briefs are often a headline and a sentence, which is a poor fit for a standalone article. Items that meet the importance threshold keep their own pages and are listed here by title.

Items with their own page

Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Model release 2026-07-21 23:16（UTC+8）

Read the full page →Read original →

OpenAI and Hugging Face partner to address security incident during model evaluation

Policy 2026-07-21 15:00（UTC+8）

Read the full page →Read original →

David Vélez and Robin Vince join the boards of the OpenAI Foundation and OpenAI Group PBC

Policy 2026-07-21 08:00（UTC+8）

Read the full page →Read original →

Short items

K3发布不到一周，美国封禁外国开源AI模型的呼声再起

News 2026-07-21 17:56（UTC+8）

K3发布不到一周，美国封禁外国开源AI模型的呼声再起麻省理工科技评论

Items with their own page

Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

OpenAI and Hugging Face partner to address security incident during model evaluation

David Vélez and Robin Vince join the boards of the OpenAI Foundation and OpenAI Group PBC

Short items

K3发布不到一周，美国封禁外国开源AI模型的呼声再起

阿里千问发布 Qwen-Image-3.0 图像生成基础模型：落字成画，字字如印

阿里千问发布 Qwen-Image-3.0，超长指令让复杂图文一键生成

Google releases series of new cheaper Gemini models

OpenAI says Hugging Face was breached by its pre-release models

实探月之暗面Kimi：K3之后，算力、开源与商业化怎么走？

阿里发布Qwen-Image-3.0图像模型，支持超长指令复杂图文生成

花旗：阿里发布通义千问3.8 Max预览版 参数2.4万亿将开源

阿里千问发布图像生成模型Qwen-Image-3.0：支持4.5k token长输入

Kimi负责人回应三大争议：SOTA模型应有自己的定价权

“Kimi时刻”反而利好芯片？华尔街：“杰文斯悖论”将再度应验！

Kimi紧急叫停新订阅：模型效率提升，算力缺口却在扩大

国产大模型Kimi K3登顶榜单，马斯克都成为背景板

国产开源大模型Kimi K3到底强在哪？何时向新用户开放？

AI“斯普特尼克时刻”！Kimi K3震动硅谷，Stability AI联合创始人：推理成本还有10至50倍下降空间

马斯克“宣战”Kimi？月之暗面回应

马斯克杠上Kimi启示录：大模型哪有护城河，但模型企业有

热点回应丨国产开源大模型参数规模冲击新高度

七月起，大模型进入「无限战争」

磐石·科学基础大模型2.0发布

花旗：阿里发布通义千问3.8 Max预览版参数2.4万亿将开源