DeepSeek puts its flagship V3 model in your pocket for free
DeepSeek just launched its official mobile app on the iOS App Store and major Android marketplaces, giving users free access to its 671-billion-parameter DeepSeek-V3 model. The app supports web search, deep thinking mode, and file and image scanning. It basically turns one of open source's hottest models into a free, pocket-sized ChatGPT rival.
After shaking up the open-source AI community with its ultra-efficient model, DeepSeek is now putting it straight into everyone's pocket.
Chinese AI startup DeepSeek officially launched its mobile app across the Apple App Store and major Android platforms, including Huawei, Xiaomi, Honor, and OPPO.
DeepSeek-V3 goes mobile
The free-to-use app gives everyday users direct access to the 6710B-parameter DeepSeek-V3 model.
Feature-wise, the app supports web search and a deep thinking mode. It also includes document and image uploads, capable of scanning and extracting text on the fly.
DeepSeek first open-sourced DeepSeek-V3 in December 2024, revealing that it trained the frontier model using just 2,048 Nvidia H800 GPUs over two months for roughly $5.58 million, delivering performance near GPT-4o. Now, users can skip third-party web clients and API configurations entirely.
Why it matters
We see this as DeepSeek's official pivot from a developer darling to a mainstream consumer challenger.
While the startup proved that frontier models can be built at a fraction of Silicon Valley's budget, the average user doesn't touch raw weights or API keys. Packing full web-search and reasoning features into a free native app brings DeepSeek into direct rivalry with mobile mainstays like ChatGPT.
The real test now shifts to inference costs: maintaining a free, fast consumer app under mass adoption could quickly test the economic limits of the budget-conscious startup.
Sources (2)
- techinasia.com(opens in a new tab)Originaltechinasia.com
- sohu.com(opens in a new tab)Reportsohu.com