Futurism

American AI Industry Trembles as Deepseek Prepares to Release New Model

When Chinese AI company DeepSeek released its cheap and serviceable V3 model early last year, it sent shockwaves throughout Silicon Valley and beyond, roiling the stock market, shaking political ...
Reuters

OpenAI says China's DeepSeek trained its AI by distilling US models, memo shows

Feb 12 (Reuters) - OpenAI has warned U.S. lawmakers that Chinese artificial intelligence startup DeepSeek is targeting the ChatGPT maker and the nation's leading AI companies to replicate models and ...
来自MSN

DeepSeek或春节发布新模型? 新论文直指大模型短板，V3.2表现亮眼

近日，科技领域传来新动态，DeepSeek正对其网页端和APP进行新功能测试，此次测试聚焦于全新的长文本模型结构，该结构可支持高达1M的上下文。不过，其API服务目前依旧保持为V3.2版本，仅支持128K上下文。 这一动作引发外界诸多猜测，不少人认为DeepSeek或许会在 ...
scmp.com

DeepSeek boosts AI model with 10-fold token addition as Zhipu AI unveils GLM-5

Chinese artificial intelligence start-up DeepSeek has updated its flagship AI model, adding support for a large context window with more up-to-date knowledge and fuelling further anticipation over its ...
cio.economictimes.indiatimes

OpenAI says China's DeepSeek trained its AI by distilling US models

OpenAI has warned US lawmakers that Chinese artificial intelligence startup DeepSeek is targeting the ChatGPT maker and the nation's leading AI companies to replicate models and ‌use them for ⁠its own ...
The Straits Times

OpenAI accuses DeepSeek of unfairly extracting results from US AI models to gain an edge

San Francisco – OpenAI claims that its Chinese rival DeepSeek is using unfair and increasingly sophisticated methods to extract results from leading US artificial intelligence (AI) models to train the ...
新浪网

DeepSeek 灰度测试百万 Token 上下文，知识库更新至 2025 年 5 月

IT之家 2 月 11 日消息，今日有多个用户反馈，称 DeepSeek 更新之后开始灰度测试最高 1M（百万）Token 的上下文长度；去年 8 月发布的 DeepSeek V3.1 上下文长度拓展至 128K。 IT之家询问 DeepSeek 后，无论是 App 还是网页端均回答“1M”上下文。 与此同时，最新 DeepSeek 的 ...
腾讯网

回顾一下DeepSeek这一年，V4发布前我们经历了什么

DeepSeek 这一年正式更新了7次，今年多了很多新模型，但性价比之王DeepSeek还是铁打的划算。2月11日，DeepSeek新版本灰度，百万上下文，知识也更新到了2025年5月，会是V4吗？时间线回到 2025，12月1日，DeepSeek-V3.2：达到了GPT-5的水平，稍低于Gemini-3.0-Pro，比起Kimi-K2 ...
至顶头条 on MSN

DeepSeek V4还没出，但它可能改变一切判断：2026年AI技术流派的完整硬 ...

之前发了很多期商业方向的博客了，其实我也是一个技术博主来着，而且研究AI，不研究技术，属于水中捞月。所以，本期我来一份硬核技术分享。
腾讯网

DeepSeek官方披露正测试新模型结构，或将在今年春节再次“炸场 ...

2月13日，澎湃新闻记者获悉， DeepSeek 网页/APP正在测试新的长文本模型结构，支持 1M上下文。 其API服务不变，仍为V3.2，仅支持128K上下文。 这也被外界认为，DeepSeek或将在今年春节再次“炸场”发布新模型，复刻去年春节现象级轰动。
新浪网

DeepSeek春节再扔王炸？官方披露正测试新模型结构

2月13日，澎湃新闻记者获悉，DeepSeek 网页/ APP 正在测试新的长文本模型结构，支持1M 上下文。其API 服务不变，仍为 V3.2，仅支持128K 上下文。 官方披露正测试新模型结构（资料图） 这也被外界认为，DeepSeek或将在今年春节再次“炸场”发布新模型，复刻去年春节 ...
U.S. News & World Report

OpenAI Says China's DeepSeek Trained Its AI by Distilling US Models, Memo Shows

Feb 12 (Reuters) - ⁠OpenAI ⁠has warned U.S. lawmakers ⁠that Chinese artificial intelligence startup DeepSeek is targeting the ChatGPT maker and the nation's leading AI companies to replicate models ...