NEWSLETTER
How the Depseek's budgetary model is compared to Chatgpt, Claude and Gemini in SQL, EDA and automatic learning
The price of Nvidia's shares fell more than 15% on Monday, January 27, after a Chinese startup, Depseek, launched its new ai model. The model The performance is on par with chatgpt, llama and claude but to a fraction of the cost. <a target="_blank" class="af ob" href="https://www.wired.com/story/openai-ceo-sam-altman-the-age-of-giant-ai-models-is-already-over/” rel=”noopener ugc nofollow” target=”_blank”>According to WiredOperai spent more than USD $ 100 million to train GPT-4. But Deepseek's V3 model was trained for only $ 5.6 million. This profitability efficiency is also reflected in the costs of the API: for every 1 million tokens, the Deepseek-Chaat model (V3) costs $ 0.14, and the Deepseek-Razoner model (R1) costs only $ 0.55 (Deepseek API prices). Meanwhile, the GPT-4O API costs $ 2.50 / 1m input tokens, and the API O1 costs $ 15.00 / 1m input tokens (Operai fire prices).
Always intrigued by the emerging LLMs and their application in data science, I decided to put Deepseek to the test. My goal was to see how well your chatbot model (V3) could help or even replace data scientists in their daily tasks. I used the same criteria of my series of previous articles, where I evaluated the performance of ChatgPT-4O vs. Claude 3.5 Sonnet vs. Advanced gemini SQL consultations, Exploratory Data Analysis (EDA)and <a target="_blank" class="af ob" href="https://towardsdatascience.com/chatgpt-vs-claude-vs-gemini-for-data-analysis-part-3-best-ai-assistant-for-machine-learning-a2078793e4fa” rel=”noopener” target=”_blank”>Automatic learning (ML).
