AITOPAI热报
Industry

Hugging Face 开发者突破 2000 万，托管数据达 300PB

Official accountClement Delangue@ClementDelangue
Post
View the post
Why we picked it

Hugging Face 用户破 2000 万、托管量到 300PB，一周新增 9PB，10 月 DAU 和收入都是历史最高。做开源模型或 agent 训练数据的人可以看看这个生态的体量，自己放数据集时也心里有数了。

Hugging Face 公布了两组新数据：平台上的 AI 开发者数量突破 2000 万，Hub 上托管的模型、数据集、agent traces 和 RL 环境总量达到 300PB。CEO Clement Delangue 称仅上周用户就新增了 9PB 内容，开源社区的产出速度可见一斑。他还透露 2025 年 10 月是 Hugging Face 历史上 DAU、MAU 和收入都最高的一个月，开源 AI 生态的商业化和用户增长同时处于高位。

Body

从内容结构看，300PB 中包含的 agent traces 和 RL 环境说明 Hub 的定位已不只是模型和数据集的托管站，而是向 agent 训练数据的存储层扩展。上周 9PB 的新增意味着用户贡献的内容约占存储增量的主力，社区驱动的数据积累仍是 Hugging Face 区别于闭源平台的核心壁垒。

Original · Clement Delangue

We just crossed: - 20M AI builders on @huggingface - 300PB of models, datasets, agent traces, and RL environments hosted on the @huggingface hub (9PB added by users just last week) Also October was our biggest month ever in terms of DAUs, MAUs and revenue. Let's go open-source AI!📊 32 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Hugging Face, 开源模型, AI开发者, 数据集.

Source attribution: curated from Clement Delangue (@ClementDelangue).

#Hugging Face#开源模型#AI开发者#数据集

Keep reading

Recent stories with the same tag or section
ProductsBuildersMulti-sourceSelected01:20
NVIDIA releases Nemotron-Personas-Japan V2 and PII Japan datasets

Fully synthetic data sidesteps the privacy-approval cycle that typically stalls Japanese-language projects, letting teams start training immediately. Model builders working on PII redaction for regulated sectors like finance and healthcare get ready-made annotations instead of hand-labelling documents. The 10-million-persona scale is rare among Japanese-language datasets, which matters for teams fine-tuning models for the Japan market rather than relying on English-centric synthetic personas.

#NVIDIA#Nemotron#synthetic data#PII detection
Event
Official accountNVIDIA AI@NVIDIAAISource
ModelsBuilders00:39
Chinese contentMiniMax 整理模型资源入口：Hugging Face 权重、API 文档与 Hailuo AI 视频平台一站汇总

MiniMax 把自家 Hugging Face 开源权重、API 文档、Hailuo 视频站和 MiniMax Code 四个入口打包发出来了。想自部署的去 Hugging Face 拉权重，想直接调 API 的看 platform.minimax.io 的指南，做视频的可以直接上 hailuoai.video。

#MiniMax#Hailuo AI#Hugging Face#开源模型
Hailuo AI@Hailuo_AISource
ModelsBuildersSelected00:36
MiniMax releases open-weight H3-Image model with 16MP output

The notable point is that 16MP output and 16-reference-image conditioning ship as open weights, not just an API feature, so teams can self-host and fine-tune instead of paying per generation. Developers running product-photography, ad-creative or character-consistency pipelines should compare it against closed APIs like OpenAI's image tools on cost and editability. The 50% launch-week discount on 16MP+ generations makes the hosted route cheap to trial before committing to infrastructure.

#MiniMax#Hailuo AI#H3-Image#open-source models
Hailuo AI@Hailuo_AISource