#DeepSee
LightNews LightNews
davidmogeladze.bsky.social
thenextweb.com/news/deepsee...
DeepSeek open-sources Huawei Ascend tools as a CUDA alternative
‘Independent, self-controlled’: DeepSeek open-sources TileLang and six modules for Huawei’s Ascend chips, offered as an alternative to Nvidia’s CUDA.
thenextweb.com
October 1, 2026 at 1:44 PM
aidailypost.com
Deepseek just rewired its open‑source AI stack for Huawei’s Ascend chips, swapping CUDA for the new TileLang. Could this be the next step toward true AGI on custom AI hardware? Dive in to see the details. #Deepseek #AscendChips #TileLang

🔗 aidailypost.com/news/deepsee...
September 30, 2026 at 3:07 PM
stephenlb.bsky.social
we've been testing models for daily use. DeepSeek-v4.1-flash for common daily tasks was the best value blocks.ai/blog/deepsee...
September 29, 2026 at 10:30 PM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: Modern Pricing, by Byrne Hobart https://www.thediff.co/archive/modern-pricing/
Teaser: Pricing is a window into strategy. In this edition of The Diff, Byrne Hobart examines how commoditization, risk allocation, and supply chain realities are reshaping the way businesses set value. It’s not just about what you charge—it’s about the market conditions you navigate and the risks you choose to own. A deep dive into the mechanics of modern pricing that leaves the obvious conclusions behind. teaser for https://www.thediff.co/archive/modern-pricing/ generated by deepseek/deepseek-v4-flash
September 29, 2026 at 4:20 PM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: Who Is Really Rich in America? (Hint: Main Street Millionaires), by Nick Maggiulli https://ofdollarsanddata.com/who-is-really-rich-in-america/
Teaser: Forget the Forbes 400. The real wealth in America isn’t held by tech founders in Silicon Valley, but by the “Main Street Millionaires” who own your local law firm, auto dealership, or oil and gas operation. Nick Maggiulli dives into new research from economists Owen Zidar and Eric Zwick to reveal who these people are, how they built their fortunes, and what it takes to join their ranks. teaser for https://ofdollarsanddata.com/who-is-really-rich-in-america/ generated by deepseek/deepseek-v4-flash
September 29, 2026 at 12:36 PM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: Guns still going strong, by Unknown http://volewica.blogspot.com/2026/09/guns-still-going-strong.html
Teaser: An anonymous cartoon about the NRA’s staying power opens this post, and the author isn’t sure who drew it. From there, it’s a pointed look at how guns and lobbying remain woven into American politics—and how little that seems to change. The tone is wry, the observation sharp, and the question of who’s really driving the debate lingers. Short, blunt, and worth reading to see where it lands. teaser for http://volewica.blogspot.com/2026/09/guns-still-going-strong.html generated by deepseek/deepseek-v4-flash
September 29, 2026 at 10:09 AM
propjerry.bsky.social
DeepSeek Infrastructure and Bridge360 Metatheory Governance Alignment: Audio overview applying Bridge360 Metatheory Model lens

#AIGovernance
#AISafety

www.pinoytoolbox.org/post/deepsee...
September 25, 2026 at 4:32 PM
propjerry.bsky.social
On DeepSeek arXiv paper "DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale": Applying Bridge360 Metatheory Model lens

#AIGovernance
#AISafety
#MachineLearning
#ReinforcementLearning
#DeepLearning

agericomontecillodevilla.substack.com/p/on-deepsee...
On DeepSeek arXiv paper "DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale"
Applying Bridge360 Metatheory Model lens
agericomontecillodevilla.substack.com
September 25, 2026 at 2:52 PM
davidmogeladze.bsky.social
thenextweb.com/news/deepsee...
DeepSeek’s annualised revenue run rate reaches $1bn, report says
DeepSeek’s annualised revenue run rate has doubled to $1bn in a few months after an API price rise, The Information reports.
thenextweb.com
September 24, 2026 at 1:35 PM
redhotcyber.bsky.social
DeepSeek V4.1 Flash ha un asso nella manica! Il nuovo modello usa 4 volte meno memoria dei precedenti

📌 Link all'articolo : www.redhotcyber.com/post/deepsee...

Luigi Zullo

#redhotcyber #cybersecurity #cybercrime #hacking #cti #ai #privacy #news #technology
September 20, 2026 at 7:24 AM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: The Price of Affirmative Action in Canadian Academia, by Quillette https://quillette.com/2026/09/16/the-price-of-affirmative-action-in-canadian-academia/
Teaser: Eric Kaufmann joins Jonathan Kay to break down his new Macdonald-Laurier Institute report on the Canada Research Chairs Program, using h-Index data to quantify what diversity quotas cost research excellence. The findings are stark—but as Kaufmann notes, university administrators remain wedded to euphemisms like "inclusive excellence," and reform won't come from within. teaser for https://quillette.com/2026/09/16/the-price-of-affirmative-action-in-canadian-academia/ generated by deepseek/deepseek-v4-flash
September 17, 2026 at 1:23 AM
monsuta.bsky.social
enclave.ai/blog/deepsee...
Enclave: DeepSeek V4.1 Flash is Now Our Best Hacking Model
DeepSeek’s 11/11 result showed why advanced agent benchmarks need to check both the outcome and the attack path: our audit confirmed six planned exploits and found five unexpected routes.
enclave.ai
September 16, 2026 at 5:40 PM
alexmilla.bsky.social
DeepSeek-V4.1-Flash Explained: The Causal Encoder-Decoder Architecture That Cuts KV Cache by 8x

miraflow.ai/blog/deepsee...
DeepSeek-V4.1-Flash Explained: The Causal Encoder-Decoder Architecture That Cuts KV Cache by 8x
DeepSeek-V4.1-Flash introduces a new Causal Encoder-Decoder architecture that shrinks KV cache storage 8x. Here is how the 552B MoE model actually works, benchmarked, priced, and called.
miraflow.ai
September 16, 2026 at 2:31 PM
bsky.shibadogcap.com
RT @voluntas: shiguredo/deepseek-harness: DeepSeek Harness: Everything is a Plugin. github.com/shiguredo/deepsee… DSH の日本語＆フォント指定対応版です。

Original: https://x.com/4ba_ba_baba/status/2099854339376562560

(リンク: https://github.com/shiguredo/deepseek-harness)
September 15, 2026 at 1:46 PM
sosadformycountry.bsky.social
Hundreds of AI agents broke into 395 organisations in 48 countries, and some went off script

www.wionews.com/world/hundre...

Download the WION App now:
onelink.to/jksrch

-Shared via WION App
Hundreds of AI agents broke into 395 organisations in 48 countries, and some went off script
GreyNoise says an attacker used hundreds of AI agents to exploit PaperCut flaws at 395 organisations in 48 countries — 11 of them in 26 seconds. The agents ran on OpenAI's Codex harness with a DeepSee...
www.wionews.com
September 14, 2026 at 2:13 PM
aidailypost.com
DeepSeek‑V4.1‑Flash slashes memory footprints using FP4 KV caching—open‑source magic that lets you run bigger models on less RAM. Curious how it works? Dive into the details and see the benchmarks. #DeepSeek #FP4 #KVCaching

🔗 aidailypost.com/news/deepsee...
September 14, 2026 at 12:24 PM
chat-o.bsky.social
DeepSeek v4.1 is on Chat-O. We used it to help us improve our photo editor, and it's now available as a model for everyone on Chat-O! chat-o.chat/blog/deepsee...
We Tested DeepSeek V4.1 Flash on Our Own Chat-O Codebase, Then Gave It Our Photo Editor · Chat-O AI
We wanted to learn how DeepSeek V4.1 Flash handles real software work, not a toy prompt. First, we handed it a coupled dependency upgrade in our Elixir and Phoenix LiveView app. It changed the…
chat-o.chat
September 12, 2026 at 2:19 PM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: Friday discussion post, by Halina Bennet https://www.slowboring.com/p/friday-discussion-post-178
Teaser: Halina Bennet was too young to remember 9/11, but like all Americans, its shadow has shaped her world. In this Friday discussion post, she shares the media she returns to each year on the anniversary—pieces she read and watched today in an effort to grasp the events she never witnessed firsthand. A personal, generational reflection on remembrance and the stories we turn to when memory isn't our own. teaser for https://www.slowboring.com/p/friday-discussion-post-178 generated by deepseek/deepseek-v4-flash
September 11, 2026 at 11:20 PM
ai-health.bsky.social
Jeudi, Anthropic a publié un rapport détaillant ce qu’elle qualifie d’"attaques de distillation" perpétrées par trois acteurs chinois – Alibaba, Moonshot AI et DeepSeek.

👉 Lire l'article complet 👇
#IA #AI
Anthropic dénonce des campagnes de distillation menées par Alibaba, Moonshot AI et DeepSee
Anthropic accuse Alibaba, Moonshot AI et DeepSeek de mener des attaques de distillation, révélant une escalade des tensions dans la course aux modèles…
aihealth.fr
September 11, 2026 at 2:23 PM
aidailypost.com
Deepseek just dropped V4.1‑Flash, a multimodal AI that cuts KV‑cache memory by 437×. Imagine long‑context agents running on a single GPU. Curious? Check out the full breakdown! #Deepseek #V4_1Flash #KVCache

🔗 aidailypost.com/news/deepsee...
September 10, 2026 at 1:41 PM
aidailypost.com
DeepSeek‑V4.1‑Flash pushes 1M context with KV‑cache tricks, sparse attention & multimodal tokens. Open weights + vLLM turn it into a transformer playground. Curious? Dive into the details! #DeepSeekV4_1_Flash #1MContext #SparseAttention

🔗 aidailypost.com/news/deepsee...
September 10, 2026 at 7:42 AM
yicaiglobal.bsky.social
DeepSeek’s New Model Shows Significantly Improved Performance in Internal Beta Testing www.yicaiglobal.com/news/deepsee...
DeepSeek’s New Model Shows Significantly Improved Performance in Internal Beta Testing
www.yicaiglobal.com
September 9, 2026 at 8:00 AM
masuda-bot.jiskey.dev.ap.brid.gy
新しい記事が投稿されました

ハドラーの魔王軍に例えると、現在の中国AI界は以下のように綺麗に分類できます。
1. 百獣魔団（クロコダ...
https://anond.hatelabo.jp/20260906021759
ハドラーの魔王軍に例えると、現在の中国AI界は以下のように綺麗に分類できます。
1. 百獣魔団（クロコダイン） ＝ Qwen（アリババ） 特徴: 圧倒的なパワーと組織力を持つ、魔王軍の「主力中の主力」。 実態: アリババという中国EC・クラウドの巨人がバックにいる、最も規模が大きく盤石な「ドコモ」的AIです。 2. 不死騎団（ヒュンケル） ＝ DeepSee…
anond.hatelabo.jp
September 5, 2026 at 5:19 PM
some-rss-feeds.bsky.social
[some-subscribed-rss] New Post: Coalescence or ... no-alescence, by Halina Bennet https://www.slowboring.com/p/coalescence-or-no-alescence
Teaser: In Texas, the most dramatic Senate race of the cycle is boiling over: both nominees have gone sharply negative, abandoned by members of their own parties. Meanwhile, up in Michigan, something entirely different is happening—candidates are drawing support from unlikely places. What explains the divergence? Slow Boring’s Halina Bennet digs into the contrasting dynamics of coalescence and its absence. teaser for https://www.slowboring.com/p/coalescence-or-no-alescence generated by deepseek/deepseek-v4-flash
September 2, 2026 at 11:10 PM
groundlabs.bsky.social
Data sovereignty is moving higher up the EU policy agenda. We explored this issue last year, showing organizations need a clear understanding of the data they hold before they can make informed decisions about where that data is allowed to go: groundlabs.com/blog/deepsee...
DeepSeek, GenAI and Data Sovereignty Explained | Ground Labs
Explore the implications of DeepSeek and GenAI on data sovereignty, plus tips for the safe use of AI for businesses
groundlabs.com
August 31, 2026 at 8:30 AM