Posts
News
Latest
Hot
On February 27, Eastern Time, OpenAI dropped a heavy bomb—GPT 4.5 research preview version is online! It is said to be their largest and most knowledgeable model, with emotional intelligence improved compared to previous versions. Boss Sam Altman stated that he would invest tens of thousands of dollars in GPUs to support it. Even more exciting is that GPT 5 is already on the way and will meet everyone in a few months. Meanwhile, the three giants DeepSeekR1, Grok3, and GPT 4.5 are taking turns to appear, the AI circle is simply bustling. So the question arises, who will win this AI race? And how can we benefit from related Concept stocks? Let’s discuss it again!
In the showdown among the three AI giants, which one is more worth being Bullish on?
DeepSeekR1: The open-source giant, low-key yet substantial.
Let’s talk about DeepSeekR1, which was developed by the Chinese team DeepSeek, focusing on reasoning abilities, particularly excelling in hardcore fields like mathematics and programming. It is said to have achieved a score of 79.8% on the AIME 2024 test, and an impressive 97.3% on the MATH-500, crushing 96% of human contestants in the Codeforces ranking. What's even more impressive is that it is open-source, available for free, with remarkably low costs – training the V3 model only cost 6 million dollars, saving 90% compared to OpenAI! It has partnered with AMD, using their GPUs and Software optimizations, offering exceptional cost performance. However, it currently seems more like a toy for academia and developers, as its commercialization has not yet fully rolled out, and its future remains to be seen.
In the showdown among the three AI giants, which one is more worth being Bullish on?
DeepSeekR1: The open-source giant, low-key yet substantial.
Let’s talk about DeepSeekR1, which was developed by the Chinese team DeepSeek, focusing on reasoning abilities, particularly excelling in hardcore fields like mathematics and programming. It is said to have achieved a score of 79.8% on the AIME 2024 test, and an impressive 97.3% on the MATH-500, crushing 96% of human contestants in the Codeforces ranking. What's even more impressive is that it is open-source, available for free, with remarkably low costs – training the V3 model only cost 6 million dollars, saving 90% compared to OpenAI! It has partnered with AMD, using their GPUs and Software optimizations, offering exceptional cost performance. However, it currently seems more like a toy for academia and developers, as its commercialization has not yet fully rolled out, and its future remains to be seen.


+2
83
5
22
Unlock Pro Investors’ Money-Making Secrets
Join Futubull Community! Now Connect Directly with Top Investors & Public Company Executives