![\"DeepSeek](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/DeepSeek-Effect.jpg\")
<\/p>\n<\/p>\n
Gone are the days of proprietary software and technologies. As AI becomes more democratized, open-source models are gaining momentum. You must have heard of DeepSeek\u2014a Chinese AGI (artificial general intelligence) startup that has made it to the headlines in the last few weeks. It\u2019s everywhere, from social media to developer community forums.<\/p>\n
\n![\"DeepSeek-R1](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/DeepSeek-R1-is-now-live.png\")
<\/p>\n
Pushing boundaries in deep learning (DL) and knowledge discovery, DeepSeek\u2019s LLM is more than just another addition to the open-source AI ecosystem. It has challenged other proprietary models by providing 100% transparency at just a fraction of the budget.<\/p>\n
What has added more to this buzz is the fact that it has been developed by a Chinese<\/i> startup (DeepSeek Artificial Intelligence Co., Ltd.). With offerings and capabilities comparable to leading Western models like OpenAI’s ChatGPT, it has captured significant attention, caused people to doubt the actual cost of AI\/ML development, and raised concerns over U.S. AI supremacy. This has paved the way for more affordable and accessible AI. Let\u2019s explore what this development has to offer and whether it is an improvement over existing AI market leaders like ChatGPT.<\/p>\n Visibly, DeepSeek has sparked significant conversations in the AI community.<\/p>\n As per Microsoft\u2019s CEO Satya Nadela<\/b>, people should have an optimistic view of this development. He wrote, \u201cJevons paradox strikes again\u2014As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can\u2019t get enough of.\u201d<\/p>\n Similarly, former Intel CEO Pat Gelsinger<\/b> sees DeepSeek as a reminder of computing\u2019s evolution, emphasizing that cheaper AI will drive broader adoption, constraints fuel innovation (Chinese engineers worked with limited computing power), and most importantly, \u201copen wins\u201d\u2014challenging the increasingly closed AI ecosystem.<\/p>\n However, not everyone is convinced. Elon Musk<\/b> and Scale AI\u2019s Alexandr Wang<\/b> remain skeptical, questioning whether DeepSeek\u2019s claims about building a competitive model with minimal computing resources can genuinely be validated. The debate will continue for some time, at least till we find answers to:<\/p>\n DeepSeek is built on a Mixture-of-Experts (MoE) architecture. In this neural network design, numerous expert models (sub-networks) handle different tasks\/tokens, but only selective ones are activated (using gating mechanisms) at a time based on the input.<\/p>\n The latest model, DeepSeek V3, has been trained on 671 billion<\/a> parameters with 37 billion<\/a> activated per token.<\/p>\n It underwent pre-training on a vast dataset of 14.8 trillion tokens, encompassing multiple languages with a focus on English and Chinese. Subsequent supervised fine-tuning (SFT) was conducted on 1.5 million samples, covering both reasoning (math, programming, logic) and non-reasoning tasks. Reinforcement learning was also applied to enhance the model’s reasoning capabilities. (source<\/a>)<\/p>\n DeepSeek (V3 in particular) tops the leaderboard of all open-source LLMs for the following benchmarks:<\/p>\n\nTable of Content<\/b><\/div>\n\n
\n
\n
<\/span>The DeepSeek Hype: Why is Everyone Talking About it?<\/h2>\n
\n
\n\n\n<\/span>Inside DeepSeek: What Powers This Open-Source LLM<\/h2>\n
<\/span>Architecture<\/h3>\n
<\/span>Model Parameters<\/h3>\n
<\/span>Training<\/h3>\n
<\/span>Evaluation Benchmarks<\/h3>\n
![\"DeepSeek](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/DeepSeek-V3-vs.-ChatGPT-4-Turbo.jpg\")
![\"ChatGPT](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/ChatGPT-Can-Be-Optimized.jpg\")
<\/a><\/p>\n![\"DeepSeek](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/DeepSeek-V3-vs.-Claude-3.5-Sonnet.jpg\")
![\"DeepSeek](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/DeepSeek-V3-vs.-Gemini-2.0.jpg\")
![\"Need](\"https:\/\/www.technoscore.com\/blog\/wp-content\/uploads\/2025\/02\/Need-Help-Deciding-Which-AI-Model-Suits-the-Best-for-your-Business.jpg\")
<\/a><\/p>\n