Search results for DeepSeek V4
Industry
U.S. prosecutors call Huawei a \'criminal enterprise\' as trial focuses on trade secrets, Iran sanctions
U.S. prosecutors told a federal court in Brooklyn that Huawei was a “criminal enterprise,” accusing the Chinese telecoms equipment maker of stealing trade secrets, violating Iran sanctions and committing financial fraud. A Justice Department prosecutor said Huawei extracted technology from U.S. companies over about 20 years and abused the financial system. Huawei rejected the allegations, saying prosecutors are linking separate matters into an organised conspiracy.
AI & Enterprise
DeepSeek V4.1 Flash nears GPT-6 Astra performance at 1.4 percent of cost
DeepSeek\'s new V4.1 Flash model posted performance close to OpenAI\'s GPT-6 Astra in a practical design-task evaluation while sharply cutting costs, results showed. OpenDesign Arena gave V4.1 Flash 81.2 out of 100, compared with 82.7 for GPT-6 Astra. The cost per finished design was $0.023 for V4.1 Flash versus $1.61 for GPT-6 Astra, and average completion time was 5.3 minutes versus 11.1 minutes.
AI & Enterprise
Value-for-money AI models reshape spending, frontier model sales wobble
Lower-cost AI models are increasingly influencing corporate spending on high-end offerings from Anthropic and OpenAI, a Financial Times report said, citing Ramp’s analysis of spending data from 70,000 companies. Anthropic’s top-tier Fable 5 accounted for about 11 percent of spending on Anthropic tools more than two months after launch. Analysts cited price and sufficient performance of older models. The shift may affect frontier AI developers investing billions of dollars in larger models.
-
AI & Enterprise
DeepSeek releases official V4 Pro model; 46 times cheaper than Claude with 5 percent performance gap
-
AI & Enterprise
Why Nvidia wants to develop even large AI models itself
-
AI & Enterprise
Local PC LLMs gain traction as tech firms step up competition
-
AI & Enterprise
China\'s top AI still dependent on Nvidia as shift to domestic chips slows
-
AI & Enterprise
ByteDance training AI model with up to 10 trillion parameters, report says
-
AI & Enterprise
Tencent opens core Hy3 AI model free globally, first since hiring ex-OpenAI researcher
-
AI & Enterprise
U.S. open-weight AI startups emerge, but checking China is unclear
-
AI & Enterprise
Four teams in Korea\'s sovereign AI foundation model project ready for second review with different scale-up strategies
-
AI & Enterprise
China AI narrows performance gap, cuts inference costs to as little as one-hundredth
-
AI & Enterprise
Poolside unveils \'Laguna S 2.1\' open-weight model for coding agents
-
AI & Enterprise
U.S. weighs curbs on Chinese open-source AI models after debut of free Kimi
-
AI & Enterprise
AI price war heats up as Altman says OpenAI can offer one-quarter of Anthropic price
-
AI & Enterprise
China AI token use surges 1,000-fold in two years, expands into Western markets
-
AI & Enterprise
OpenAI launches new flagship model GPT-5.6 Sol with midrange pricing, top-tier performance
-
AI & Enterprise
U.S. tech firms increasingly use Chinese open-source AI models