DeepSeek develops AI models that developers access through an API, and it also runs a free consumer chatbot. The Chinese lab is reportedly aiming to close a $7.5 billion second financing by the end of October, according to The Information.

The paid API, not the chatbot, supplies most of the revenue. The Next Web, citing The Information, put DeepSeek’s annualized revenue pace at $1 billion, up from below $500 million a few months earlier. That figure extrapolates recent sales over a year; it isn’t $1 billion already collected.

Recent pricing helps explain the jump. Fees rose 2.3 to 4.5 times depending on the model, The Next Web reported, while founder Liang Wenfeng told investors the customer base held steady. DeepSeek hasn’t commented on the revenue number, so both the sales pace and retention claim remain reported rather than company-confirmed.

The distinction between its products is practical. A chatbot user visits DeepSeek’s free consumer service; an API customer pays to call the models from other software. With the chatbot free, the paid API is what converts model use into revenue for DeepSeek.

DeepSeek only began taking outside money this year. The Next Web reported that its first round closed in June at roughly $7 billion, with Liang providing a large share. A successful end-October close would put a similarly sized second round less than five months after the first; for now, the $7.5 billion remains a fundraising target.