While the Chinese startup’s latest version V4 lags behind top models like GPT-5.4 and Gemini 3.1 Pro Preview in financial, legal and coding tasks, it features a new attention mechanism that reduces the computational cost
DeepSeek launched V4, its new open-source foundation model, in Pro (1.6T total/49B active parameters) and Flash (284B/13B) versions, both with 1M-token context window and up to 90% compute reduction via Sparse Attention. Despite trailing Claude Opus 4.6 and Gemini 3.1 Pro in financial/legal benchmarks (63.87% accuracy), V4 keeps DeepSeek competitive in math/STEM/coding while rumors of a $20B+ Tencent/Alibaba investment signal growing pressure on Western AI stacks.
00:00/00:00
您的浏览器不支持 audio 标签。
Listen to this article
1x
Chinese artificial intelligence (AI) startup DeepSeek launched its highly anticipated V4 open-source foundational model on Friday, upping the ante in a race with global peers to develop a technology that underpins data centers and AI applications.
DeepSeek unveils next-gen AI model as Huawei vows ‘full support’ with new chips
China's DeepSeek unveils long-awaited V4 AI model
As agentic AI pushes rivals to raise prices and cap usage, Deepseek V4 is a good-enough model for almost nothing
China's DeepSeek releases preview of long-awaited V4 model as AI race intensifies
Three reasons why DeepSeek’s new model matters
China’s DeepSeek unveils latest models a year after upending global tech
DeepSeek presenta V4, e rilancia la sfida sull’IA
DeepSeek-V4 preview now available with open-source access · TechNode
US-China AI race intensifies as DeepSeek releases 'reduced' cost model
Underwhelming or underrated? DeepSeek V4 shows “impressive” gains
The company says its cost-efficient new V4 model is competitive with top closed-source models from OpenAI and Google DeepMind.
Chinese AI darling DeepSeek has launched its long-awaited V4 LLM in preview, as speculation around a possible first funding round…
Chinese AI lab Deepseek has released V4-Pro and V4-Flash, two new models with up to 1.6 trillion parameters and a…
Chinese AI startup DeepSeek has released a preview version of its long awaited V4 large language model.
The long-awaited V4 is more efficient and a win for Chinese chipmakers.
The Chinese start-up has introduced only a few incremental updates in recent months, while competitors have released new models.
