The artificial intelligence (AI) world saw a critical change in 2025 when DeepSeek launched as a new Chinese startup that started commanding attention after its debut in 2023. This piece examines DeepSeek, from its operation, key facts, major data, and its comparison to OpenAI, updated for 2026 with the current V4 models, user and traffic figures, and API pricing (checked on October 5, 2026).

What is DeepSeek?

  1. Founding and Leadership: DeepSeek is an open-source large language model development company created by Liang Wenfeng in 2023. (Source: Reuters)
  2. Company Headquarters: The Chinese artificial intelligence company operates under the official name Hangzhou DeepSeek Artificial Intelligence Co., Ltd. and maintains headquarters in Hangzhou China. (Source: Wikipedia)
  3. Company Size: The company has achieved important milestones with a small staff of about 160 employees as of 2025. (Source, accessed October 2026: Wikipedia)

Key DeepSeek statistics

  1. Efficient Development: The development pace for the DeepSeek-R1 AI model reached two months and its creation cost less than 6 million dollars through the deployment of Nvidia H800 chips. The frugal innovation framework led to quick and budget-friendly development thus earning widespread praise. (Source: Reuters)
  2. App Store Success: DeepSeek launched its first free chatbot application which supported both iOS and Android operating systems. The free application surpassed ChatGPT and became the most popular download available through the U.S. iOS App Store on January 27, 2025. (Source: The Atlantic)
  3. Global Impact: Several analysts recognize DeepSeek as an important influence that can threaten American AI models in their worldwide dominance. (Source: Reuters)
  4. Open-Source Approach: DeepSeek makes open-source models available for unrestricted modification and unrestricted use while many U.S. companies maintain proprietary control over their models. Its R1, V3.1, V3.2 and V4 models were released under the MIT License. (Source, accessed October 2026: Wikipedia)
  5. Market Reaction: The R1 model’s success at DeepSeek caused Nvidia and other tech companies to witness substantial drops in their stock prices because of the disruptive potential of DeepSeek’s technology. (Source: Reuters)

2026 update – current models: DeepSeek launched its V4 generation (V4-Pro and V4-Flash) on April 24, 2026, and retired the older deepseek-chat and deepseek-reasoner model names on July 24, 2026. V4-Pro became generally available on August 13, 2026, and DeepSeek-V4.1-Flash, with native multimodal support, followed on September 10, 2026. As of October 5, 2026, the API offers two models, deepseek-v4-pro and deepseek-flash, each with a 1M-token context window. (Sources: DeepSeek API change log and DeepSeek Models & Pricing, both accessed October 5, 2026)

DeepSeek registered user statistics

  1. While there are no official data, Deepseek has approximately 5-6 million users worldwide during January 2025 as estimated through the company’s 2.6 million app acquisitions and 7 million daily Google query peak. (Source: SEO AI) Later figures are far higher: DeepSeek reached 130 million active users at the end of 2025, up from 33 million in January 2025 (Source: Business of Apps, updated July 16, 2026), and it had 130 million monthly active users in China alone as of May 2026, according to QuestMobile data cited by About Chromebooks (published August 15, 2026). The same article cites Similarweb data showing 346.4 million visits to deepseek.com in July 2026, down 7.25% month over month, with China (49.27%), Russia (9.15%) and the United States (6.08%) sending the most traffic.

DeepSeek download statistics

  1. The download numbers for Deep Seek exceeded 10 million by January 2025. (Source: Google Play) In total, the app has been downloaded 173 million times since its launch in January 2025, although quarterly downloads fell from 83 million in Q1 2025 to 13 million in Q1 2026. (Source: Business of Apps, updated July 16, 2026)
  2. In the weeks after launch in early 2025, nearly 40% of DeepSeek downloads came from China while the United States represented 16%. For 2025 as a whole, Business of Apps (updated July 16, 2026) puts China at 32%, Russia at 9%, India at 6% and the United States at 5% of downloads. (Source: Business of Apps)
DeepSeek downloads by country statistics

DeepSeek downloads by country statistics

CountryDownloads, early 2025 (%)
China39%
United States16%
India10%
South Korea7%
Russia6%
France4%
United Kingdom3%
Brazil3%
Germany3%

Source: Business of Apps

DeepSeek vs OpenAI Comparison

  1. Development Cost: The cost of developing DeepSeek-R1 reached under $6 million but OpenAI’s GPT-4 development needed around $100 million. (Source: Reuters)
  2. Performance Benchmarks: At its launch in January 2025, DeepSeek-R1 achieved better results than OpenAI’s then-current o1 model (o1-1217) during two of five coding tests where it led in LiveCodeBench (65.9% vs. 63.4%) and SWE Verified (49.2% vs. 48.9%). (Source: Demandsage)
  3. Market Impact: DeepSeek-R1’s introduction caused major tech corporations to suffer substantial stock slide which demonstrates the disruptive nature of this innovative technology. (Source: The Atlantic)
  4. The open-source DeepSeek-V3 model contains 671B parameters which uses Mixture-of-Experts architecture and activates 37B parameters per token. It implements new load distribution methods coupled with multiple token prediction capabilities while using 14.8T tokens for training. V3 has since been succeeded by the V4 family, launched on April 24, 2026. (Source: Docsbot)

2026 pricing comparison with OpenAI’s current models: the table shows standard API list prices in US dollars per 1 million tokens, as published on October 5, 2026. DeepSeek charges peak and off-peak rates (off-peak is half the peak rate); OpenAI prices are the standard, short-context rates.

ModelInput (per 1M tokens)Output (per 1M tokens)
DeepSeek V4-Pro (deepseek-v4-pro)$1.32 peak / $0.66 off-peak$3.96 peak / $1.98 off-peak
DeepSeek V4.1-Flash (deepseek-flash)$0.30 peak / $0.15 off-peak$1.20 peak / $0.60 off-peak
OpenAI GPT-6 Astra$10.00$50.00
OpenAI GPT-6.1 Sol$2.00$10.00
OpenAI GPT-6 Luna$0.10$0.50

Sources: DeepSeek Models & Pricing (cache-miss input prices) and OpenAI API pricing, both accessed October 5, 2026. On list price, DeepSeek’s top model costs a fraction of OpenAI’s flagship, while OpenAI’s smallest model, GPT-6 Luna, is cheaper than DeepSeek’s Flash model. Price says nothing about quality, so test both on your own tasks.

Conclusion

The year 2025 established DeepSeek as an industry-changing technology within artificial intelligence. The open-source framework, budget-friendly development, and exceptional operation of DeepSeek establish it as an industrial disruptor while bringing new possibilities for modern innovation. In 2026 it followed up with the V4 model family and reached 130 million monthly active users in China (QuestMobile, May 2026).

By the numbers: On the day DeepSeek’s assistant overtook ChatGPT in Apple’s app store, investors wiped $592.7 billion off Nvidia’s market value, a record one-day loss for a Wall Street stock, according to Reuters.

The platform allows users to observe features simultaneously providing support through growth initiatives that define the evolution direction of AI.

Do you want to stay updated on the latest market and AI trends? Follow us and share this article on social platforms so others can discover revolutionary information!


Frequently Asked Questions

It depends on the task. DeepSeek’s API list prices are well below those of OpenAI’s flagship model (October 2026), and DeepSeek-R1 edged out OpenAI’s o1 in some coding benchmarks in early 2025, but no single model wins every test. DeepSeek’s open-source releases are a point of difference from OpenAI’s proprietary models.

DeepSeek achieves its main function through natural language processing by enabling the creation of chatbots alongside content production and programming support capabilities. The AI software enables model deployment for various fields encompassing technology and education.

DeepSeek emerges from the mind of its Chinese founder Liang Wenfeng who now directs AI innovation at the company.

DeepSeek delivers its chatbot application at no cost which makes high-level artificial intelligence capabilities obtainable to a broader spectrum of users. The company makes its models available through open-source access allowing developers to work with them.

The open and unavailable AI models from DeepSeek need basic security precautions as part of its responsible AI approach to stop destructive content generation and misuse of their platform.

Related reading