TrustFinance is trustworthy and accurate information you can rely on. If you are looking for financial business information, this is the place for you. All-in-One source for financial business information. Our priority is our reliability.

TrustFinance Global Insights
4月 24, 2026
2 min read
34

Chinese AI firm DeepSeek has released preview versions of its new flagship open-source artificial intelligence model, V4. The release includes two versions: the DeepSeek-V4-Pro with 1.6 trillion parameters and the more efficient DeepSeek-V4-Flash with 284 billion parameters. Both models feature a context window of 1 million tokens, enhancing their information processing capabilities, though they are currently limited to text.
As one of China's leading "AI Tigers," DeepSeek's latest release solidifies its position in the competitive AI landscape. The V4 models were reportedly trained on advanced AI chips from Huawei, specifically the Ascend 950, navigating around US technology restrictions on NVIDIA chips. This development comes as reports indicate Tencent and Alibaba are in talks to invest in DeepSeek at a valuation exceeding $20 billion.
According to data from the company, the V4-Pro model achieves top-tier performance on coding benchmarks, matching OpenAI’s GPT-4.5 and closely trailing models from Google and Anthropic. This advancement in open-source AI challenges the dominance of closed-source rivals and could influence market dynamics by offering comparable performance, potentially with greater efficiency and accessibility.
The launch of DeepSeek V4 marks a significant milestone for open-source AI, intensifying competition among global tech giants. Market observers will closely monitor the model's adoption rates and the outcome of the company's high-stakes investment talks, which could further accelerate its growth.
Q: What are the new DeepSeek V4 models?
A: The company released two models: DeepSeek-V4-Pro, a large model with 1.6 trillion parameters, and DeepSeek-V4-Flash, a smaller, more efficient version with 284 billion parameters.
Q: How does DeepSeek V4's performance compare to others?
A: The V4-Pro model's performance on coding benchmarks is comparable to leading models like OpenAI's GPT-4.5, making it a strong competitor in the AI industry.
Source: Investing.com

TrustFinance Global Insights
AI-assisted editorial team by TrustFinance curating reliable financial and economic news from verified global sources.
Related Articles

15 5月 2026
Poland's WIG30 Index Closes Down 2.01%