Nvidia Eyes $2.5 Billion Investment in Thinking Machines Lab 
ByteDance Secures $29.6 Billion Loan to Boost AI Investment 
Thomson Reuters C-Track Cyberattack Hits Courts Across U.S. and Canada 
OpenAI Launches GPT-6 Astra With Advanced AI Agent Capabilities 
South Korea, US Discuss Chip Investments Amid Tariff Plans 
Spire Healthcare Shares Jump on £1.03 Billion Takeover Deal 
Tesla Cybercab Launch Puts Robotaxi Ambitions in Focus 
Lynas Rare Earths Shares Rise After Takeover Talks Revealed 
GLP-1 Weight-Loss Drug Use Surges Among U.S. Children 
Xiaomi Shares Jump on Europe EV Expansion Plan 
Apple’s Phil Schiller Steps Back as Leadership Shake-Up Accelerates 
SpaceX Turbine Push Unlikely to Threaten Howmet, Bernstein Says 
Amazon Prime Air Cargo Plane Crash Kills Five at Miami Airport 
Huawei Launches Mate XT2 Foldable Phone in China 
Microsoft to Reveal Azure Revenue in Major Reporting Shake-Up 
SEC Seeks ISS Client Voting Records in Proxy Adviser Probe 
Add as a preferred source on Google
DeepSeek, a fast-rising Chinese AI startup, has unveiled a preview of its highly anticipated V4 model, marking a major step toward China’s technological independence. The new model, optimized for Huawei’s Ascend chips, highlights a strategic shift away from Nvidia hardware and reflects China’s push to build a self-sufficient artificial intelligence ecosystem.
According to DeepSeek, the Pro version of V4 performs exceptionally well in global knowledge benchmarks, ranking just behind Google’s Gemini-Pro-3.1 among leading models. This achievement positions DeepSeek as a serious contender in the global AI race, especially within the open-source community. The model has already gained rapid traction, topping trending charts on Hugging Face, a key platform for machine learning developers.
The collaboration between DeepSeek and Huawei is particularly significant. Huawei confirmed that its chips were used during parts of V4’s training process, reinforcing confidence in domestic semiconductor capabilities. Analysts note that Huawei’s Ascend chips are currently China’s strongest alternative to Nvidia, especially as U.S. export restrictions continue to limit access to advanced American hardware.
DeepSeek V4 stands out for its ability to process extremely long and complex text inputs, supporting over one million tokens—comparable to top-tier models like GPT-5.4 and Claude Opus 4.6. Despite this capability, it operates at a fraction of the computational cost, making it an attractive option for developers and enterprises. However, it currently lacks multimodal support, meaning it cannot process images or video.
The launch comes amid rising geopolitical tensions, with U.S. officials raising concerns over intellectual property and data security. Several Western governments have already restricted the use of DeepSeek’s technology. Meanwhile, competition within China is intensifying, with rivals such as Zhipu AI and MiniMax reacting strongly to the V4 release.
DeepSeek is positioning V4 for advanced AI agent applications, which require higher computational efficiency. While early benchmarks are promising, experts suggest waiting for broader real-world testing. As open-source models continue to close the gap with proprietary systems, DeepSeek’s latest innovation could reshape the global AI landscape.
