✨DeepSeek, a China-based AI startup, recently launched its latest AI model, DeepSeek-R1, which has attracted widespread attention in the tech industry. The model is claimed to be able to match, or even surpass, the performance of leading models such as OpenAI's ChatGPT o1 in some aspects, while offering significantly lower development and operational costs.
DeepSeek-R1 is an open-source model that leverages reinforcement learning technology to improve its reasoning capabilities. It uses widely available datasets, which can include data from various sources both within China and abroad. Because the model is open-source, it can leverage existing datasets in the community, which are not necessarily limited to one specific country.
The DeepSeek R1 is built on the Nvidia H800 chip, which is a variant of the Nvidia H100 chip designed specifically for the Chinese market with lower interchip bandwidth. Although Nvidia is an American company, the H800 chip was specifically designed to comply with US export restrictions on China.
This model competes directly with flagship products from OpenAI and Google, showing that innovation in AI development does not always come at a high cost.
DeepSeek-R1 was trained at a cost of about $5.6 million, which is much lower than the typical development costs incurred by AI companies in the US.
Training and development costs:
🔹 OpenAI's GPT-4: $78 million for training and development.
🔹 Google Gemini Ultra: $191 million for development and training.
🔹 OpenAI tends to rely on large investments to achieve high model quality, primarily through extensive training data and intensive computing.
🔹 Google, with Gemini, is relying not only on financial resources but also on integration with Google's extensive ecosystem to increase the functionality and accessibility of AI models.
💡 The launch of DeepSeek R1 has sparked significant volatility in the cryptocurrency market. Investors have begun taking profits or selling assets due to uncertainty about the impact of this innovation on the overall tech economy. Bitcoin and altcoins such as Render (RNDR), Near Protocol (NEAR), The Graph (GRT), and Artificial Superintelligence Alliance (FET) have seen their value plummet, with some tokens dropping between 7-9%, and Node.AI (GPU) dropping as much as 20% in a single day.📊
There is a very large possibility that the proceeds from the liquidation of these technology stocks will go into $BTC