Chinese AI research lab DeepSeek has garnered global attention with the release of its open-source AI model, DeepSeek-R1. The model promises significant improvements in areas such as mathematical reasoning, code generation, and cost efficiency, marking DeepSeek as a potential rival to industry giants like OpenAI. The financial markets reacted strongly to this development, as reported by Moneycontrol.
US stock futures dropped sharply, with Nasdaq contracts declining 1.9% during Asian trading hours, amid fears that DeepSeek’s affordable AI model could challenge the business models of major US firms like Nvidia, OpenAI, and Google. The ripple effect extended to Japanese chipmaker stocks, with shares of Nvidia supplier Advantest Corp. falling 8.6 per cent in Tokyo.
ALSO READ | OpenAI Chief Praises DeepSeek’s R1 Model, Elon Musk Takes A Dig: Here’s What Industry Leaders Have To Say
Many are wondering what it is about this thing that has sent ripples across markets. Let us go over the top 10 things that we know about DeepSeek R1 model.
Top 10 Things That We Know About DeepSeek R1 Model
- Cost-effective: One of the prime attraction points of the DeepSeek R1 model is its pricing. At $0.55 per million input tokens, it is significantly cheaper than OpenAI’s models, making it an attractive option for developers and organisations.
- Hybrid architecture: Built on hybrid architecture, DeepSeek R1 model combines large-scale reinforcement learning (RL) and chain-of-thought reasoning to enhance response accuracy. It comes in two versions- DeepSeek R1 and DeepSeek R1 Zero, with the latter excelling in unsupervised fine-tuning.
- Rapid Development: DeepSeek developed R1 in just two months with an investment of $6 million, a fraction of the billions spent by OpenAI, Google and Microsoft on their AI models.
- Exceptional Performance: DeepSeek R1 matches or outperforms OpenAI’s models in benchmarks like mathematics, coding and reasoning tasks. It boasts a 97 per cent success rate in coding tasks, surpassing OpenAI’s o1 in some areas.
- Compact Versions: DeepSeek has released six versions of R1, optimised to run efficiently on laptops. One of these models outperforms OpenAI’s o1 mini in specific benchmarks.
- Open-Source Initiatives: DeepSeek’s open-source approach aligns with China’s growing trend of adopting open-source practices in AI development, fostering innovation and collaboration.
- Global Recognition: Microsoft CEO Satya Nadella has acknowledged DeepSeek’s potential, emphasising the need to take China’s AI advancements seriously.
- Social Media Buzz: Users have praised DeepSeek R1 for its seamless performance on smartphones and superior results in tasks like mathematical computation and visual demonstrations.
- Top Downloaded App On Apple App Store: DeepSeek AI app has become the most downloaded app on Apple Play Store in US and has beaten the ChatGPT app.
- Tech Titans React: OpenAI chief Sam Altman has praised the performance of DeepSeek R1. X owner Elon Musk took a dig at it over data privacy concerns. Many industry leaders have reacted to it.
