DeepSeek reportedly gets China's approval to buy NVIDIA's H200 AI chips

Mariella Moon / engadget - The Chinese government has given DeepSeek its approval to purchase NVIDIA’s H200 AI chips, according to Reuters. ByteDance, Alibaba and Tencent have also reportedly received permission from Beijing to buy a total of 400,000 H200 GPUs. Reuters says Chinese…

Back to Top / Friday, January 30, 2026, 8:21 am / permalink 18562 / 4 stories in 5 wks

DeepSeek to Unfurl AI Model with Enhanced Coding Capabilities in February: Report

Jon Swartz / techstrong - DeepSeek is preparing to launch its next-generation artificial intelligence (AI) model, V4, in mid-February, according to a report in The Information, citing two people with direct knowledge of the company's plans. The upcoming release marks the Chinese s…

Back to Top / Friday, January 9, 2026, 12:20 pm / permalink 17684 / 5 stories in 1 month

DeepSeek Kicks Off the Year With New Paper Introducing the mHC Architecture

pandaily - DeepSeek has unveiled a new research paper proposing mHC, a novel architecture designed to stabilize large-scale model training while preserving performance gains.

Back to Top / Thursday, January 1, 2026, 11:20 pm / permalink 17414 / 9 stories in 2 months

Xiaomi Rolls Out MiMo-V2-Flash to Rival DeepSeek, Moonshot, and OpenAI

Maxwell Mutuma / parameter - TLDR Xiaomi unveils MiMo-V2-Flash, an open-source AI model designed to compete with DeepSeek, Moonshot AI, and OpenAI. The model is a step toward artificial general intelligence (AGI), focusing on reasoning, coding, and problem-solving tasks. MiMo-V2-Flas…

Back to Top / Wednesday, December 17, 2025, 7:21 am / permalink 17063 / 3 stories in 2 months

DeepSeek Accused of Using Smuggled NVIDIA Chips: Report

Jon Swartz / techstrong - Chinese artificial intelligence (AI) startup DeepSeek allegedly relied on banned NVIDIA Corp. semiconductors to develop its latest AI model, according to a report from The Information that has sparked denials from the chipmaker and raised questions about …

Back to Top / Wednesday, December 10, 2025, 11:20 am / permalink 16786 / 6 stories in 2 months

DeepSeek's Math Model Beats Humans at Their Own Game, Then Gives Away the Playbook

Robert Brown / implicator - DeepSeek's new math model matches OpenAI and Google's gold-medal IMO performance, then does what they won't: releases everything. The 685B-parameter system beat the best human Putnam score by 28 points. But did it reason, or just remember?

Back to Top / Saturday, November 29, 2025, 7:20 pm / permalink 16409 / 6 stories in 3 months

DeepSeekMath-V2 Matches OpenAI and Google with IMO Gold Medal Win

Markus Kasanmascheff / winbuzzer - DeepSeek has released DeepSeekMath-V2, an open-weight AI that has matched OpenAI and Google's Gold Medal performance at the IMO 2025.The post DeepSeekMath-V2 Matches OpenAI and Google with IMO Gold Medal Win appeared first on WinBuzzer.

Back to Top / Thursday, November 27, 2025, 5:20 pm / permalink 16359 / 3 stories in 3 months

Hurricane Melissa Has Meteorologists Terrified

Molly Taft / wired - The storm, which is set to make landfall in Jamaica Tuesday, has stunned meteorologists with its intensity and the speed at which it built.

Back to Top / Tuesday, October 28, 2025, 11:21 am / permalink 15490 / 10 stories in 4 months

New Deepseek model drastically reduces resource usage by converting text and documents into images — 'vision-text compression' uses up to 20 times fewer tokens

tomshardware - Developers of the Chinese Deepseek artificial intelligence have found a novel way to reduce the number of tokens it uses, particularly when accessing memories. By converting blocks of text into images and leveraging visual processing rather than text pars…

Back to Top / Tuesday, October 21, 2025, 9:21 am / permalink 15269 / 3 stories in 4 months

OpenAI Blocks Users From Making AI Videos of Martin Luther King Jr.

Bruce Gil / gizmodo - The company says it’s adding new “guardrails” to protect historical figures from AI misuse.

Back to Top / Friday, October 17, 2025, 10:22 am / permalink 15189 / 9 stories in 4 months

OpenAI Blocks Users From Making AI Videos of Martin Luther King Jr.

Bruce Gil / gizmodo - The company says it’s adding new “guardrails” to protect historical figures from AI misuse.

Back to Top / Friday, October 17, 2025, 10:20 am / permalink 15183 / 0 stories in 4 months

A first look at Sora 2, OpenAI’s wild new video platform

Hal Crawford / mumbrella - When GPT5 was released a couple of months ago, it felt like a letdown. Where was the staggering new capability, the “wow” moment that knocked everyone for six? Perhaps, many wondered, AI was finally starting to slow down? Fear not (or stay afraid) — the p…

Back to Top / Friday, October 3, 2025, 8:21 am / permalink 14973 / 9 stories in 5 months

Sorry, but DeepSeek didn’t really train its flagship model for $294,000

Tobias Mann / theregister - Training costs detailed in R1 training report don't include 2.79 million GPU hours that laid its foundation Chinese AI darling DeepSeek's now infamous R1 research report was published in the Journal Nature this week, alongside new information on the compu…

Back to Top / Friday, September 19, 2025, 1:21 pm / permalink 14744 / 2 stories in 5 months

DeepSeek AI's code quality depends on who it's for (and China's opinion of them)

techspot - According to a report by The Washington Post, CrowdStrike carried out an experiment in which the security firm asked DeepSeek English-language prompt requests for help writing programs.Read Entire Article

Back to Top / Thursday, September 18, 2025, 1:21 pm / permalink 14707 / 4 stories in 5 months

DeepSeek Is Working on an AI Agent: Will It Be Better Than ChatGPT?

Katelyn Chedraoui / cnet - The Chinese AI company's reasoning model supercharged the global AI race. An agentic upgrade could have a similar effect.

Back to Top / Thursday, September 4, 2025, 3:21 pm / permalink 13824 / 1 stories in 6 months

DeepSeek’s V3.1 goes open, targets parity at a fraction of the cost

Marcus Schuler / implicator - DeepSeek's V3.1 delivers Claude-level coding performance at 1/70th the cost through open source. The Chinese startup's MIT-licensed model challenges American AI economics while optimizing for domestic chips—signaling parallel ecosystems.

Back to Top / Thursday, August 21, 2025, 7:20 am / permalink 12780 / 4 stories in 6 months