DeepSeek nears first funding round as investors scramble for stakes
Media reports on May 22 said the latest round had expanded to 70 billion yuan ($10.33 billion), with a pre-money valuation of about $45 billion.
Media reports on May 22 said the latest round had expanded to 70 billion yuan ($10.33 billion), with a pre-money valuation of about $45 billion.
At the new rates, DeepSeek said usage cost of its V4-Pro model now drops to industry-low levels, with cached input costing as little as 2.5 fen, the smallest unit of the Chinese currency, per million tokens.
DeepSeek’s image understanding feature began a phased rollout in late April 2026 and was made widely available to users on May 9, according to the company.
This comes as competition among the country’s large language model developers intensifies, driving demand for capital to secure compute, attract talent and accelerate commercialization.
The release follows an extended, almost 3-month delay that industry sources attribute in part to a transition in training infrastructure from Nvidia GPUs to Huawei Ascend chips.
The city trails only Beijing in total computing capacity, according to industry data, reflecting the capital’s concentration of national-level data centers and research institutions.
The update marks the first time DeepSeek has introduced functional tiering since its debut in November 2023, signaling a shift toward more refined product positioning.
Most current AI agents lack long-term memory, which can lead to context loss in extended conversations or multi-session tasks, forcing users to repeatedly explain preferences and instructions.
The rapid integration reflects Alibaba’s strategy of pairing foundation models with deeply embedded enterprise scenarios, Wukong’s spokesperson said.
The factory is seen as a cornerstone for positioning Jinhua as a hub for AI terminal manufacturing while strengthening the Yangtze River Delta’s domestically controlled tech supply chain.