DeepSeek raises API pricing up to 1,100%, reopens $8bn round, invests in robotics

DeepSeek raises API pricing for its V4 models

BEIJING, Aug 13 (Reuters) - Chinese artificial intelligence startup DeepSeek will raise API prices for its V4-Pro and V4-Flash models and introduce ⁠peak and off-peak pricing, a company statement on Thursday showed. The new rates - ranging from 50% to 1,100% above current ⁠prices, depending on the model, token type, and time ⁠of use - will take effect on August ⁠17, according to the statement. Reporting ⁠by Xiuhao Chen and Ryan Woo; Editing by Emelia Sithole-Matarise Our Standards: The Thomson Reuters Trust Principles., opens new tab

DeepSeek raises $8bn and buys into a robot maker

DeepSeek built its name making AI cheap. This week it hiked prices, reopened an $8bn round, and bought a robot-maker stake. DeepSeek spent early 2025 proving that a good AI model did not have to cost a fortune. This week it has been busy proving it can spend one. In a handful of days, the Chinese lab reopened a multi-billion-dollar fundraising. It took a stake in the country's best-known robot maker. And it told customers it would charge them more. An $8bn round, reopened DeepSeek has restarted its second funding round, seeking close to $8bn, Bloomberg reported. The raise values the Hangzhou startup at about 500 billion yuan, or $74bn. Monolith Management, an early backer of the Chinese AI champion Moonshot, is in talks to join. The round is a restart, not a fresh start. DeepSeek paused the process last month after leaked remarks from founder Liang Wenfeng to investors caused friction. The valuation has climbed since its first external round earlier this summer. That round closed near 350 billion yuan and roughly doubled Liang's net worth to about $36bn. The company wants to raise about as much again. Discussions are ongoing, Bloomberg cautioned, and the size, timing, and investor list can still change. What the money is for marks the real shift. DeepSeek plans to spend part of it building its own data centres, led by a large facility in Inner Mongolia. The lab that made its name on efficiency now needs its own compute. On the same day the round resurfaced, it also warned users of a significant price rise. DeepSeek is quietly rewriting its own cheap-AI story. A robot bet in the same week The second move points somewhere new. DeepSeek has invested 140.8 million yuan, about $20.8m, in Unitree Robotics' Shanghai listing, Reuters reported from a stock-exchange filing. The stake buys 2.31% of the offering's strategic placement, with a 36-month lock-up. It reads as a commitment, not a trade. Alongside it sits a pact to jointly develop AI models for humanoid robots. The two Hangzhou firms will pair DeepSeek's models with Unitree's work in motion control and embodied intelligence. Each will favour the other, Unitree for training services, DeepSeek for robots. The target is the field's hardest problem. Can a robot "brain" make sense of an unfamiliar room and turn an instruction into a reliable action? The pairing is telling. DeepSeek's models are strong on coding, maths, and reasoning. But that strength sits in language, not in the multimodal systems that read the physical world. Its VL2 vision research was never folded into the flagship. A separate visual-reasoning project appeared and then vanished in April without explanation. Unitree offers the missing half, robots and the scarce physical-world data they generate. A rich price for the hardware Unitree priced its IPO on Thursday, the payoff of a listing TNW has tracked since its earlier 50-billion-yuan target. It will raise about 6.1 billion yuan, or $904m, at a valuation near $9bn, Reuters reported. That makes it the first humanoid-robot maker to list on the mainland. Backers already include vehicles tied to Tencent, Alibaba, and China Mobile. The numbers are steep. At 150.80 yuan a share the offering carries a price-to-earnings ratio of 219. The industry average is near 39, according to Chinese outlet Cailian Press. Revenue more than quadrupled to 1.7 billion yuan in 2025, and humanoids became the biggest line. But first-quarter profit, stripped of one-offs, fell 52.6% as research and marketing costs climbed. US sales, 13.3% of last year's revenue, sit exposed to fresh American curbs on Chinese robots. Put the three moves together and a pattern shows. The company that proved AI could be cheap is now raising billions, building data centres, and charging more. It is also buying into machines that move. DeepSeek is betting that the next phase will not be cheap at all. Whether the market that fell for the discount will pay the new price is the open question.

DeepSeek invests $20.8 million in Unitree's Shanghai IPO

BEIJING, Aug 6 (Reuters) - Chinese artificial intelligence startup DeepSeek has invested 140.8 million yuan ($20.8 million) in robot maker Unitree (688836.SS), opens new tab and agreed to jointly develop AI models for humanoid machines, according to a stock-exchange filing. DeepSeek received 933,399 Unitree shares, representing 2.31% of the shares allocated through the Shanghai initial public offering's strategic placement. The companies, both based in China's eastern tech hub of Hangzhou, said they would ⁠combine DeepSeek's expertise in AI models with Unitree's work in mechanical engineering, motion control and embodied intelligence. Unitree, also known as Yushu Technology, would give preference to DeepSeek when procuring model-training services and technical solutions. DeepSeek would similarly favour Unitree when purchasing robots or exploring embodied-AI applications, the filing said. The partnership targets a key bottleneck for humanoid developers: building a robot "brain" capable of understanding unfamiliar surroundings and reliably turning instructions into physical actions. China has developed relatively inexpensive robots capable of walking, running and performing choreographed routines, but useful ⁠autonomous work requires models trained on scarce physical-world data, including demonstrations of manipulating objects and responding to changing environments. Chinese robot makers are operating data-collection facilities in which humans repeatedly guide humanoids through tasks such as folding clothes and opening doors. DeepSeek's AI models have performed strongly on ⁠coding, mathematics and reasoning tests, while competing aggressively on price. But the company's strengths have been concentrated in language-model series rather than a multimodal system able to process vision alongside text. DeepSeek has ⁠released separate vision-language research models, including VL2, but they are not integrated into its flagship commercial model. A DeepSeek researcher briefly published a project in April on improving visual ⁠reasoning through points and bounding boxes before the paper and official GitHub repository were withdrawn without a public explanation. The Unitree partnership could provide DeepSeek with robots and physical-world data to accelerate such work. Reporting by Eduardo Baptista. Editing by Mark Potter Our Standards: The Thomson Reuters Trust Principles., opens new tab * Suggested Topics: * Emerging Markets Eduardo Baptista Thomson Reuters Eduardo Baptista is a Senior Correspondent for Reuters based in Beijing, covering China's technology, space, and automotive industries. He has led enterprise and investigative reporting on China's military-linked companies, artificial intelligence and semiconductor supply chains, as well as macroeconomic and industrial policy. Baptista has reported from China for nearly a decade and holds a BA in History from the University of Cambridge.

DeepSeek Harness launches as open source rival to Claude Code, alongside V4-Pro on API with higher prices

DeepSeek is expanding beyond the model layer and deeper into the software developers use to put AI agents to work. The Chinese AI lab on Thursday launched the official version of DeepSeek-V4-Pro, an updated flagship model focused heavily on agentic workloads, alongside DeepSeek Harness v0.1, a new open-source agent harness that gives developers an alternative to integrated coding-agent environments such as Anthropic's Claude Code. Together, the releases amount to a broader developer push from DeepSeek. V4-Pro is now available across DeepSeek's web interface, mobile app and API, with native support for the OpenAI Responses API and integration with Codex. DeepSeek Harness, meanwhile, is entering developer preview under the MIT license and the code is available now for download and use on GitHub. It's built around an unusually modular premise: practically every part of the agent runtime can be swapped out as a plugin. But developers accessing V4 through DeepSeek's API will soon pay considerably more for it. DeepSeek is simultaneously abandoning its existing flat API pricing in favor of peak and off-peak rates beginning at 16:00 UTC on Sunday, Aug. 16 (2 am ET). Even the discounted off-peak cache-miss and output prices will be substantially higher than the prices available today. The combination is significant because DeepSeek is no longer competing solely over model intelligence and token prices. With Harness, it is moving into the layer that determines how models use tools, manipulate files, maintain sessions and execute long-running agent workflows -- territory where Anthropic's Claude Code and other coding agents have become increasingly important developer products. DeepSeek builds its own agent harness DeepSeek describes Harness, or , as an open-source agent harness built on Cordis, a framework designed around composable plugins. Its guiding principle is simple: "Everything is a plugin." That extends to models, tools, skills, sessions, sandboxes, filesystems, loops, orchestration and user interfaces, according to DeepSeek. Rather than making those components fixed pieces of a single coding agent, Harness is designed to let developers mix, replace and extend them. The project is available under the MIT license and can currently be launched from npm with . DeepSeek also provides instructions for building it directly from source. The repository describes the software explicitly as a developer preview and warns that "THERE WILL BE COMPATIBILITY-BREAKING CHANGES." That caveat matters for enterprise developers. Harness is not yet being presented as a stable drop-in production platform. But its architecture points toward a potentially important strategy: DeepSeek can now offer developers not only models but an open framework for assembling the systems that surround them. That makes Anthropic's Claude Code and OpenAI's Codex useful competitive references, although the products should not be treated as functionally identical. DeepSeek Harness is an open-source, model-agnostic alternative to the agent infrastructure underlying Claude Code and Codex -- not yet a full replacement for either product's broader developer experience. It can already inspect repositories, edit files, execute shell commands, search files and the web, maintain plans, invoke skills, delegate work to subagents and enforce approval policies. Those are the essential capabilities that make Claude Code and Codex agentic coding tools rather than autocomplete systems. DeepSeek explicitly describes Standard mode as a full coding agent with file editing, shell access, search, planning, subagents and workflows. Its local web interface lets users select a workspace and approve sensitive operations. But Claude Code and Codex now extend well beyond that agent loop. Here's a quick comparison: DeepSeek Harness instead emphasizes modularity and replacement: the model itself is another plugin rather than necessarily the center of a vertically integrated stack. DeepSeek's repository was already attracting significant developer attention on launch day, showing roughly 27,500 GitHub stars and 2,000 forks as of Aug. 13, although those rapidly changing figures are best viewed as a snapshot rather than an adoption metric. V4-Pro gets an agent-focused upgrade Harness arrives alongside the general-availability release of DeepSeek-V4-Pro-0813. DeepSeek originally introduced the V4 family in preview in April. The lineup consists of the 1.6-trillion-parameter V4-Pro, with 49 billion parameters activated per token, and the smaller 284-billion-parameter V4-Flash, with 13 billion activated. Both support context windows of up to one million tokens. The company's Aug. 13 release therefore is not the first appearance of V4-Pro. It is the transition from the earlier preview into an updated official version, with DeepSeek emphasizing agent performance. "The official version of DeepSeek-V4-Pro has been released, featuring significantly enhanced agent capabilities and support for the Responses API and Codex integration," DeepSeek says on its API website. "It is now fully available across the web, mobile app, and API; we welcome your testing and feedback." DeepSeek's changelog similarly says the general-availability model has "significantly enhanced Agent capabilities," particularly in production environments. Developers using the API do not have to change model identifiers: now resolves to the latest V4-Pro version. The company has also added native OpenAI Responses API support, lowering the amount of integration work required for applications already built around that interface. DeepSeek says V4-Pro is optimized for OpenAI's own open source harness, Codex, with one-click setup. Its current API documentation lists Responses API, tool calling, JSON output and an Anthropic-format API among the supported interfaces for both V4-Pro and V4-Flash. For developers using DeepSeek directly rather than through an API, V4-Pro is now accessible through "Expert Mode" on the company's app and website. Reasoning effort becomes another deployment knob DeepSeek is also making reasoning effort an explicit control across V4-Pro and V4-Flash. The V4 model documentation describes three levels: Non-think, designed for fast routine tasks; Think High, intended for more complex problem-solving and planning; and Think Max, which allocates substantially more reasoning to difficult problems. That distinction can be operationally important for agent systems because maximum reasoning on every step can consume unnecessary time and tokens. A coding agent might use relatively little reasoning to inspect a file or execute a routine tool call, then increase effort when diagnosing a difficult bug or planning a multi-stage code change. DeepSeek's latest benchmark table suggests the 0813 model improves substantially on agent-oriented tests, although the figures are company-reported and some results depend on the harness configuration. DeepSeek reports V4-Pro-0813 scores of 87.9 on Terminal Bench 2.1, 74.1 on Toolathlon-Verified, 71.1 on DSBench-FullStack and 67.2 on DSBench-Hard. It does not lead every comparison in DeepSeek's own table: Fable 5, for example, scores 77.9 on Toolathlon-Verified and 77.2 on DSBench-FullStack. There is an especially important qualification buried beneath the benchmark table. For public Code Agent tasks, DeepSeek says V4-Pro-0813 was tested using its upcoming DeepSeek Harness in "minimal mode." In other words, some of the agent results arriving alongside Harness are not purely model benchmarks. They measure the model operating inside an agent execution environment -- precisely the software layer DeepSeek is now releasing to developers. A sharp reversal in DeepSeek's API price trajectory The bigger immediate change for teams already running DeepSeek in production may be pricing. DeepSeek's current API documentation lists V4-Flash at $0.14 per million cache-miss input tokens and $0.28 per million output tokens, while V4-Pro costs $0.435 for cache-miss input and $0.87 for output. Cache hits are dramatically cheaper at $0.0028 for Flash and $0.003625 for Pro. Those prices themselves represented a major reduction from V4's original April launch economics. When V4 arrived in April, V4-Pro was priced at $1.74 per million cache-miss input tokens and $3.48 per million output tokens. By late May, DeepSeek had made a 75% reduction permanent, intensifying its position as an unusually inexpensive option for high-volume agent workloads. Now the pendulum is moving in the other direction. Beginning Aug. 16 at 16:00 UTC, DeepSeek will charge different rates depending on when API calls occur. Peak hours are 01:00-04:00 UTC and 06:00-10:00 UTC (9:00 PM - 12:00 AM ET and 2:00 AM - 6:00 AM ET, respectively) with all other hours classified as off-peak. Off-peak rates are half the corresponding peak prices. For V4-Flash, off-peak cache-miss input rises from $0.14 to $0.22 per million tokens, while output rises from $0.28 to $0.66. During peak hours those rates reach $0.44 input and $1.32 output. V4-Pro moves from $0.435 per million cache-miss input tokens and $0.87 output today to $0.66 and $1.98 off-peak, respectively. Peak rates rise to $1.32 input and $3.96 output. The increases are even more pronounced for cached input. V4-Pro cache hits rise from $0.003625 per million tokens today to $0.022 off-peak and $0.044 at peak. Flash moves from $0.0028 to $0.007 off-peak and $0.014 peak. The new prices still position DeepSeek as an affordable alternative via API to Western proprietary labs, but Reuters reported Thursday that, depending on model, token category and time of use, the changes represent increases ranging from 50% to more than 1,100% over existing rates. That makes the "50% lower" off-peak framing potentially misleading without context. Off-peak is 50% cheaper than DeepSeek's new peak rate; it is not a 50% discount from the API prices developers are paying today. For a simple workload consisting of one million cache-miss input tokens plus one million output tokens, V4-Pro currently costs $1.305. The same token mix will cost $2.64 off-peak, roughly twice as much, or $5.28 during peak hours, more than four times the current price. V4-Flash moves from $0.42 under the same simple calculation to $0.88 off-peak and $1.76 peak. Actual application costs will vary considerably depending on the ratio of cached input, uncached input and generated output, making those combined figures illustrative rather than universal total-cost estimates. DeepSeek is moving up the agent stack The timing makes the strategic direction difficult to miss. When DeepSeek released the V4 preview on April 24, the major story was how much frontier-class capability the company could deliver with an unusually efficient architecture. V4-Pro uses a hybrid attention design combining Compressed Sparse Attention and Heavily Compressed Attention; at a one-million-token context, DeepSeek says it requires only 27% of the single-token inference FLOPs and 10% of the KV cache required by V3.2. By late May, the discussion had shifted toward what those efficiencies meant economically for high-volume agents, whose repeated context reads can make caching a major component of inference costs. DeepSeek's steep V4 price cuts amplified that advantage. The Aug. 13 releases move the competition another layer upward. DeepSeek now has an updated V4-Pro tuned around agent workloads, standardized interfaces designed to make it easier to connect with existing developer tooling, configurable reasoning effort, and an MIT-licensed harness for controlling the models, tools, sandboxes, filesystems and orchestration surrounding an agent. At the same time, DeepSeek is demonstrating that developers cannot assume its aggressively low API rates are permanent. For organizations considering the platform, workload scheduling, caching behavior and the option to run open weights on their own infrastructure now become more important parts of the total-cost calculation. That leaves DeepSeek pursuing two potentially conflicting advantages at once: making its agent stack more accessible and open while making its own hosted API considerably more expensive. For enterprise developers, Harness may ultimately be the more consequential part of Thursday's announcement. Models can increasingly be swapped behind standardized interfaces. The harness that controls how an agent reasons, invokes tools, edits software and persists across a workflow can be much harder to replace. DeepSeek is now competing for that layer, too.

DeepSeek raising API prices by up to 1,100% starting Aug. 16

The Chinese AI startup will introduce peak and off-peak billing for its V4-Flash and V4-Pro models, replacing a single flat rate DeepSeek is raising the prices it charges developers to access its V4-Flash and V4-Pro models, with increases ranging from 50% to more than 1,100% depending on the model, token type, and time of use. DeepSeek said the updated pricing goes live at 16:00 UTC on August 16. The price hike comes alongside a new peak and off-peak billing structure. Peak hours are defined as 01:00-04:00 and 06:00-10:00 UTC, with off-peak rates set at half the peak rates, according to DeepSeek's pricing page. For V4-Flash, output tokens will cost $1.32 per million during peak hours and $0.66 per million during off-peak hours, up from the current flat rate of $0.28 per million, the company said. V4-Pro output tokens will rise to $3.96 per million at peak and $1.98 per million off-peak, compared with the current $0.87 per million. Input token prices are also climbing. V4-Flash cache-miss input tokens will rise to $0.44 per million at peak hours, up from $0.14, while V4-Pro cache-miss input tokens will increase to $1.32 per million at peak, up from $0.435, DeepSeek said. The company said it is adjusting pricing "to allocate resources more reasonably," and that the tiered structure is intended to shift developer workloads toward less congested periods, according to Fortune. Even at the new peak rates, DeepSeek's prices remain below those of some competitors. Anthropic's Fable 5 charges $50 per million output tokens, according to Fortune. DeepSeek released V4-Flash earlier this month, with research firm Artificial Analysis finding it was the least expensive well-known AI model to run globally at the time -- costing roughly 3 cents per benchmark test. The Hangzhou-based startup had previously offered a 75% promotional discount on V4-Pro through May 5, underscoring how dramatically its pricing strategy has shifted in recent months. DeepSeek gave advance warning last week that a price increase was coming, without specifying the exact amounts. The company has also begun laying groundwork for an IPO and closed its first outside funding round, which topped $7 billion.

DeepSeek increases prices for AI services by multiple times | Fortune

DeepSeek is steeply raising the prices for its flagship V4 models, bringing the low-cost provider's rates closer to those of major artificial intelligence rivals. A new peak-hour pricing will increase the Chinese company's rates by more than four times from the current levels, according to a post on its website Thursday. The price increases will take effect on Aug. 16. Going forward, users will pay $1.32 for 1 million output tokens during peak hours, and half that during off-peak hours, for the DeepSeek-V4-Flash model. That's up from $0.28 for 1 million tokens previously. The increase still leaves DeepSeek's pricing below that of some of its main competitors. Anthropic PBC's state-of-the-art Fable 5 service sets that pricing at $50. DeepSeek gave an advance warning last week that it would hike prices, without specifying the exact increases. DeepSeek's V4-Pro model will cost $3.96 for 1 million tokens at peak hours and half that at non-peak hours, it said. That's up from the current $0.87 per million tokens. The Hangzhou-based AI lab said it's revising and adjusting the pricing "to allocate resources more reasonably." The dynamic pricing strategy is designed to encourage developers and enterprises to shift their work to less congested periods. DeepSeek's strategy has sparked a global debate in recent weeks over the cost of high-performing AI tools. Its pricing has been so far out of the usual range for AI models that it has prompted the idea of a DeepSeek "death zone," where costlier or less capable models would be obviated. It's also raised questions about the initial public offering plans of OpenAI and Anthropic because of the risks the firm poses to their profits and business models. DeepSeek is in the middle of a massive fundraising and has begun preparations for an initial public offering as soon as this year, Bloomberg has reported. Founder Liang Wenfeng will for the first time need to balance investor expectations, rapid market expansion and the intense capital demands of building out costly computing infrastructure.

DeepSeek releases official V4 Pro model as it steps up expansion

DeepSeek said the V4-Pro-0813 greatly enhances agent capabilities and is available through the API, APP and Web channels. The company also said it will raise API ⁠pricing ⁠for its V4 Pro and V4 Flash models and introduce peak and off-peak pricing. Chinese artificial intelligence startup DeepSeek on Thursday formally released its official V4 Pro model, aiming to regain ground against fast-moving domestic rivals as it expands hiring, computing capacity and fundraising efforts. DeepSeek said the V4-Pro-0813 greatly enhances agent capabilities and is available through the API, APP and Web channels. The company also said it will raise API ⁠pricing ⁠for its V4 Pro and V4 Flash models and introduce peak and off-peak pricing. The release will be closely watched after DeepSeek's cheaper V4 Flash model unexpectedly outperformed the April preview of V4 Pro in several independent tests. That was unusual because Pro is designed as the company's more capable product, suggesting DeepSeek had improved its technology rapidly between the preview release and the ⁠official launch. DeepSeek became one of China's most closely watched AI companies after its R1 model went viral in early 2025, raising questions about ⁠whether powerful AI systems could be built at much lower cost than those of U.S. rivals. Its early lead has since been challenged by a string of releases from Chinese competitors including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance. DeepSeek has also faced the harder task of turning its prominence into a lasting business. Reuters reported in July that the company was planning a new fundraising round at a valuation of about $74 billion, weeks after raising about $7.4 billion in its first outside financing round in June. The fundraising marked a change for a company ⁠that had long avoided external capital. It also reflects the growing cost of competing in AI, which requires large investments in computer chips, data centres and specialised staff. DeepSeek has said it aims to at least double staffing across departments, including data-centre and AI-agent teams. Reuters also reported in July that it had increased private hiring of chip-design engineers to develop its own AI chip, an effort that could reduce reliance on suppliers including Nvidia and Huawei.

DeepSeek Introduces Peak-Hour Pricing That Quadruples Current Levels | PYMNTS.com

The pricing adjustment announced Thursday by the artificial intelligence (AI) company will take effect on Sunday (Aug. 16), according to the report. When the changes are implemented, users of the DeepSeek-V4-Flash model will pay $1.32 for 1 million output tokens during peak hours, and half that during off-peak hours. Those figures are up from the current 28 cents for 1 million tokens, the report said. For DeepSeek's V4-Pro model, the new price will be $3.96 for 1 million tokens during peak hours and half that during non-peak hours, up from the current 87 cents per million tokens, per the report. The report said DeepSeek's new prices remain lower than those of its main competitors. Moonshot's Kimi K3 costs $15 per million output tokens, while Anthropic's Fable 5 is priced at $50, according to the report. In a Thursday update to its change log announcing the pricing adjustment, DeepSeek said: "With the official release of the DeepSeek V4 model family, we will update and adjust API pricing. To allocate resources more reasonably, we will adopt peak/off-peak pricing, with off-peak prices set at half of the peak-hour prices, encouraging users to schedule their tasks based on actual usage." DeepSeek initially gained attention in January 2025 when it debuted an AI model that sent shockwaves through the AI world by offering performance comparable to those of American rivals OpenAI and Meta while using substantially fewer Nvidia chips. It was reported in April that DeepSeek was offering developers discounts amid increasing competition in China's AI space. DeepSeek said at the time that it was offering a 75% discount to developers using its DeepSeek-V4-Pro model until May 5. On Aug. 6, it was reported that DeepSeek had resumed its second funding round and is looking to raise close to $8 billion at a valuation of about $74 billion. It was reported Wednesday (Aug. 12) that DeepSeek is building a new team to compete with Anthropic's Claude Code in the market for AI agents that automate work for business professionals. For all PYMNTS AI and digital transformation coverage, subscribe to the daily AI and Digital Transformation Newsletters.

DeepSeek Resumes Funding Round to Raise $8 Billion | PYMNTS.com

The Chinese artificial intelligence startup is raising the funds at a valuation of about $74 billion, according to the report. The discussions are ongoing, and the details could change, per the report. DeepSeek did not immediately reply to PYMNTS' request for comment. The Bloomberg report said DeepSeek is "riding high" after releasing its V4 Flash AI model, which gained attention for its cost-performance offering, just like the earlier model that made the company famous in early 2025. It was reported July 25 that DeepSeek paused its second round of funding days after viral comments on American-Chinese AI competition were widely attributed to the company's founder, Liang Wenfeng. The comments, which the report said were unverified, were said to be from a transcript of a meeting Liang held with unidentified parties and involved Liang discussing a reliance on Nvidia chips for AI development and China's ongoing lag in AI sophistication compared to the U.S. It was reported less than two weeks earlier, on July 14, that DeepSeek was considering a second financing round weeks after completing its first. DeepSeek wrapped its first round of funding near the end of May, raising $7 billion at a $52 billion valuation. The July 14 report said DeepSeek had begun preliminary discussions with new investors about another funding round that would value the company at roughly $71 billion before the deal. That figure would be a 37% increase in valuation. DeepSeek's unusually fast funding schedule was related to expectations of increased capital needed to construct its own data center and acquire more chips, according to the report. It was reported July 15 that DeepSeek had built an annualized revenue base approaching $500 million and that this had strengthened the company's plans to raise billions of dollars and pursue a public listing. DeepSeek's annualized revenue reportedly had reached between $400 million and $500 million, driven largely by sales of cloud-based access to its models through application programming interfaces. DeepSeek initially gained attention in January 2025 when it debuted an AI model that sent shockwaves through the AI world by offering performance comparable to those of American rivals OpenAI and Meta while using substantially fewer Nvidia chips.

DeepSeek reopens $8 billion funding round: takes stake in Unitree

DeepSeek has reopened a funding round that aims to bring in nearly $8 billion at an estimated $74 billion valuation. The move comes just days after the Chinese AI company warned that its model prices are about to jump sharply. At the same time, DeepSeek picked up a 2.31% stake in humanoid robot maker Unitree, paying 140.8 million yuan ($20.8 million) through Unitree's Shanghai IPO placement. That puts DeepSeek on a bigger path than cheap language models alone. It's stepping into robotics, and Unitree just arrived with momentum: a Shanghai STAR Market IPO that valued the company at close to $9 billion, raised about $904 million, and made it mainland China's first publicly listed humanoid robot company. The two companies said they'll work together on humanoid robot models, mixing DeepSeek's LLMs with Unitree's robotics and motion-control systems. The goal is straightforward enough: help robots understand instructions better, react in real time, and take on more complicated tasks. Under the deal, Unitree also gets priority access to DeepSeek's training services, while DeepSeek gets priority access to Unitree hardware. A big chunk of the new money is expected to go toward data centers and computing capacity in Inner Mongolia. That comes only days after DeepSeek signaled a steep increase in model pricing, which analysts took as a clear attempt to start turning scale into revenue. If you keep an eye on AI and robotics in China, this one matters. There's nothing for anyone to download yet. Still, if this partnership actually clicks, DeepSeek gets a live proving ground for its models, and Unitree gets a prominent partner just as the race starts shifting past chatbots and cloud software into embodied intelligence.

DeepSeek releases official V4 Pro model as it steps up expansion

BEIJING, Aug 13 (Reuters) - Chinese artificial intelligence startup DeepSeek on Thursday formally released its official V4 Pro model, aiming to regain ground against fast-moving domestic rivals as it expands hiring, computing capacity and fundraising efforts. DeepSeek said the V4-Pro-0813 greatly enhances agent capabilities and is available through the API, APP and Web channels. The company also said it will raise API pricing for its V4 Pro and V4 Flash models and introduce peak and off-peak pricing. The release will be closely watched after DeepSeek's cheaper V4 Flash model unexpectedly outperformed the April preview of V4 Pro in several independent tests. That was unusual because Pro is designed as the company's more capable product, suggesting DeepSeek had improved its technology rapidly between the preview release and the official launch. DeepSeek became one of China's most closely watched AI companies after its R1 model went viral in early 2025, raising questions about whether powerful AI systems could be built at much lower cost than those of U.S. rivals. Its early lead has since been challenged by a string of releases from Chinese competitors including Moonshot AI, Zhipu AI, MiniMax, Alibaba and ByteDance. DeepSeek has also faced the harder task of turning its prominence into a lasting business. Reuters reported in July that the company was planning a new fundraising round at a valuation of about $74 billion, weeks after raising about $7.4 billion in its first outside financing round in June. The fundraising marked a change for a company that had long avoided external capital. It also reflects the growing cost of competing in AI, which requires large investments in computer chips, data centres and specialised staff. DeepSeek has said it aims to at least double staffing across departments, including data-centre and AI-agent teams. Reuters also reported in July that it had increased private hiring of chip-design engineers to develop its own AI chip, an effort that could reduce reliance on suppliers including Nvidia and Huawei. (Reporting by Eduardo Baptista; Editing by Kirsten Donovan)