DeepSeek slashes V4 Pro developer pricing
Chinese AI lab DeepSeek is aggressively cutting prices for its newly released flagship model, intensifying competition across the country's artificial intelligence sector. The Hangzhou-headquartered laboratory is offering a 75 per cent discount to developers using DeepSeek V4 Pro, released the previous week after months of anticipation. DeepSeek is also reducing fees for input cache hits across its family of AI platforms to a tenth of their original pricing, sharply lowering costs for frequent users sending similar or repeated requests. The promotion comes as OpenAI, Anthropic and Alphabet's Google rapidly push new releases that can be expensive to access, with Chinese AI companies discounting to incentivise users to switch.
V4 Pro features and ecosystem integrations
DeepSeek V4 Pro carries a one-million-token context window that enables processing of complex code bases and long documents, according to coverage published on July 26. The model supports integration with Claude Code, OpenClaw and OpenCode, easing work within the broader AI ecosystem. Akshar Keremane, co-founder of Bangalore-based AI start-up O-Health, said the combination of pricing, open-source availability and the one-million-token context window lowers barriers for developers, start-ups and small enterprises and allows experimentation at capability and scale that was not previously available. The move threatens to reignite a price war that erupted after DeepSeek upended the industry with the R1 in 2025.
HONOR YOYO first to integrate DeepSeek-V4 on Android
Honor device's AI assistant YOYO has become the first Android platform to integrate the DeepSeek-V4 large model, according to a July 26 product announcement. The integration places the newly released V4 family on a mainstream consumer hardware assistant alongside the broader pricing push.
Second funding round reportedly suspended
DeepSeek has reportedly paused its second funding round after a leaked recording of founder Liang Wenfeng's investor meeting circulated online. Bloomberg reported on July 25 that DeepSeek verbally informed potential second-round investors that agreements would not be signed on schedule in the coming days, citing people familiar with the matter. The people said the funding process may resume later. Part of the reason for the pause, according to those cited, was Liang's displeasure at the leaked transcript, in which he spent more than three hours outlining the company's vision, its commitment to artificial general intelligence, and the gap between the United States and China in AI. The authenticity of the leaked transcript has not been verified, and DeepSeek did not immediately respond to a Bloomberg email seeking comment on the transcript or the funding.
Capital plans and IPO pipeline
DeepSeek completed its first funding round in June 2026, raising US$7 billion, and had originally planned to raise at least 10 billion yuan in the second round, with a target pre-money valuation of at least 480 billion yuan — a substantial increase over the roughly US$50 billion first-round valuation, according to earlier Bloomberg reporting cited by Lianhe Zaobao. The company has also reportedly begun preparing for an initial public offering that could see an application filed as early as this year. DeepSeek executives have told potential investors the company will prioritise breakthrough AI research over short-term commercial returns, and Liang has pledged to continue developing open-source large models and pursue artificial general intelligence.
What to watch next
DeepSeek may revisit the second-round funding process in the coming weeks, while the company is also reportedly preparing an IPO filing that could come as early as 2026. Honor's YOYO deployment of DeepSeek-V4 will serve as the first consumer test of the new model on Android hardware, and competitors are expected to respond with their own pricing moves as the Chinese AI price war reignites.
Share this article
