DeepSeek’s New AI Model Is by Far the Cheapest of Well-Known Models to Run, Research Firm Says

A DeepSeek AI sign is seen at a building where the Chinese start-up's office is located in Beijing, China, February 19, 2025. (Reuters)
A DeepSeek AI sign is seen at a building where the Chinese start-up's office is located in Beijing, China, February 19, 2025. (Reuters)
TT

DeepSeek’s New AI Model Is by Far the Cheapest of Well-Known Models to Run, Research Firm Says

A DeepSeek AI sign is seen at a building where the Chinese start-up's office is located in Beijing, China, February 19, 2025. (Reuters)
A DeepSeek AI sign is seen at a building where the Chinese start-up's office is located in Beijing, China, February 19, 2025. (Reuters)

A version of Chinese startup DeepSeek's flagship AI model is by far the least expensive to run on benchmark tests among well-known models globally and more than 100 times cheaper to run than Anthropic's Claude Fable 5, according to a research firm.

DeepSeek, which sources have said is preparing for a potential IPO, officially released its V4-Flash model on Friday, its latest attempt to regain momentum by doing what it is best known for - offering ultra-low-cost ‌AI alternatives.

The startup's ‌R1 model became a global sensation in early 2025, ‌triggering ⁠a selloff in ⁠global technology stocks and raising questions about the large amounts US companies were spending on AI.

DeepSeek's V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens, according to research firm Artificial Analysis. A token is a unit of data used to measure AI usage.

San Francisco-based Artificial Analysis estimated V4-Flash's average cost at 3 cents per test, compared with 86 cents for Kimi K3 from Chinese rival Moonshot AI, $1.86 for ⁠OpenAI's GPT-5.6 Sol and $3.15 for Claude Fable 5.

The comparison provides ‌a more realistic measure of value than pricing ‌alone because it accounts for the amount of data a model must process and ‌generate to complete a task. A model with low headline price can still ‌prove expensive if it requires significantly more steps to produce an answer.

DeepSeek once commanded most of the headlines about Chinese AI development but was quickly besieged by many domestic rivals including other startups such as Moonshot, MiniMax and Z.AI as well as tech giants ‌like ByteDance and Alibaba. All are vying with US tech firms for global adoption, targeting businesses seeking cheaper ways to ⁠deploy AI at ⁠scale.

Artificial Analysis said DeepSeek's V4-Flash model scored 50 out of 100 on its Intelligence Index, which combines results from nine benchmarks spanning coding, reasoning and workplace-style assignments.

That's the same score as Google's Gemini 3.6 Flash, and one point behind Meta's Muse Spark 1.1 and GLM-5.2 from Z.AI which is also known as Zhipu.

Moonshot's Kimi K3, however, scored a 57 while Anthropic's Claude Opus 5, Fable 5, and OpenAI GPT-5.6 scored nine or more points higher.

DeepSeek is also preparing a more powerful version of its model, called the V4-Pro. It has not given a date for that version's official release.

Separately on Monday, Alibaba unveiled its largest and most capable artificial-intelligence model to date, the Qwen3.8-Max, which is not far behind in size when compared with an offering from domestic rival Moonshot AI launched last month.



China’s Lenovo Posts 43% Jump in Q1 Revenue, Highest in Five Years

The Lenovo logo is seen in this illustration photo January 22, 2018. (Reuters)
The Lenovo logo is seen in this illustration photo January 22, 2018. (Reuters)
TT

China’s Lenovo Posts 43% Jump in Q1 Revenue, Highest in Five Years

The Lenovo logo is seen in this illustration photo January 22, 2018. (Reuters)
The Lenovo logo is seen in this illustration photo January 22, 2018. (Reuters)

China's Lenovo Group reported a 43% jump in quarterly revenue on Thursday, beating forecasts and sending shares of the world's largest computer maker surging, as it rides an AI hardware boom and reaps the benefits of a global memory chip shortage.

Lenovo's revenue rose to $26.94 billion in the three months ended June 30, beating analyst expectations of $22.3 billion, as the consumer electronics hardware giant benefited from artificial intelligence-driven demand and solid PC sales.

It was ‌the group's highest ‌quarterly revenue growth in the last five years, as ‌AI-related ⁠revenue grew 60% ⁠year-on-year to $9.3 billion, accounting for 35% of total revenue in its fiscal first quarter.

The company swung to a net loss attributable to shareholders of $609 million from a profit of $505 million last year, compared to the average analyst estimate of $589 million profit, according to data compiled by LSEG.

The company said the loss was primarily due to a non-cash fair value loss of US$1.7 billion arising ⁠from the revaluation of warrants issued in 2025.

Lenovo's shares hit ‌an all-time high on Thursday before ‌the results announcement, bringing its year-to-date gains to 225%. The shares surged as much as ‌17% after the results announcement.

Its US competitors Dell, Hewlett Packard and Super ‌Micro have been some of Wall Street's best performers this year but have raised prices by 10% to 30% due to soaring NAND and DRAM memory chip costs.

Lenovo's PC, tablet and smartphone division, which accounted for about 64% of total revenue, ‌reported a 27% year-on-year increase in revenue during the period. Adjusted net income, which excludes one-off items and non-cash charges, ⁠more than doubled ⁠to $1.075 billion.

R&D expenses jumped 30% year-on-year, the company said.

Global PC shipments declined by 2% year-on-year in the second quarter of 2026 to 16.6 million units for the first time since Q1 2025 due to memory-driven cost pressures, according to Counterpoint Research.

Lenovo retained its market lead in the second quarter, giving it a market share of 25.6%. Its AI server pipeline reached $54.0 billion, up 157% quarter-over-quarter, reflecting demand from hyperscalers, AI cloud and enterprise AI clients, its earnings report said.

Lenovo's strong performance comes after the company warned earlier this year of pressure on PC shipments as the industry grapples with a memory chip shortage that is getting more severe.

It has also raised PC prices to mitigate the impact of soaring memory costs.


Brazil Demands Safety Controls on Discord After Livestreamed Suicide

The Discord logo is seen in this illustration taken November 7, 2022. (Reuters)
The Discord logo is seen in this illustration taken November 7, 2022. (Reuters)
TT

Brazil Demands Safety Controls on Discord After Livestreamed Suicide

The Discord logo is seen in this illustration taken November 7, 2022. (Reuters)
The Discord logo is seen in this illustration taken November 7, 2022. (Reuters)

Brazilian authorities are calling on US-based messaging platform Discord, popular among teens, to take action after a suicide case sparked outcry and the first lady called for its shutdown.

Investigators say a 13-year-old girl was encouraged to self-harm and take her own life during a Discord livestream, and five teenagers have been arrested in connection to the death.

The office of the Attorney General of the Union (AGU), a public body that defends the legal interests of the Brazilian state, told AFP on Tuesday it requested and has received a proposal from Discord to "adopt measures aimed at strengthening the safety of its users, particularly children and adolescents."

In a statement sent to AFP, Discord said it was "cooperating with authorities in the investigation into this serious case."

Discord said it had additionally "deactivated, before the teenager's death," a "private server" suspected of being used by "individuals involved in criminal activities encouraging self-harm."

The AGU has demanded concrete measures, like age verification of users.

In Brazil, a law that came into effect earlier this year obliges platforms to link the accounts of users under age 16 to their parents' accounts, and requires platforms to verify user age.

Following the recent death, Rosangela da Silva, wife of left-wing President Luiz Inacio Lula da Silva, called for Discord to be shut down and urged "the judiciary to remove this horrible network from the Internet."

Discord claims more than 90 million active daily users worldwide.

In 2024, the platform X was blocked for 40 days in Brazil, until the social media network owned by the world's richest man Elon Musk complied with the Supreme Court's orders to remove accounts spreading disinformation.


Google’s Gemini AI Assistant Reaches One Billion Users

The Gemini app icon in this illustration taken June 5, 2026. (Reuters)
The Gemini app icon in this illustration taken June 5, 2026. (Reuters)
TT

Google’s Gemini AI Assistant Reaches One Billion Users

The Gemini app icon in this illustration taken June 5, 2026. (Reuters)
The Gemini app icon in this illustration taken June 5, 2026. (Reuters)

Google's artificial intelligence assistant Gemini has surpassed one billion monthly users, CEO Sundar Pichai announced Tuesday, a week after a major reorganization of the tech giant's AI division.

"It's our fastest growing product ever," and the 14th product to reach a billion users, Pichai posted on social media.

The milestone, which Google calculated based on how many people use the Gemini app across its ecosystem, gives the search giant a victory amid a major overhaul of its AI division.

Demis Hassabis, who won a Nobel Prize in chemistry, recently stepped back from the day-to-day management of Google DeepMind, the lab that designs the Gemini models. And several leading researchers left as well, including Jeff Dean, a central figure in the group's engineering efforts for nearly three decades.

Its new flagship model, Gemini 3.5 Pro, initially promised for June, still has not been released, while its rivals have launched several models that are considered more powerful, particularly for generating computer code -- currently the most lucrative area in the sector.

In the face of this technological lag, Google's audience remains its main asset as it integrates AI into its ecosystem, from search and Gmail to Google Maps and YouTube.

OpenAI reached one billion users across all its interfaces, led by ChatGPT, the lab announced in July without specifying whether that figure was monthly or weekly users.

The San Francisco-based AI lab sparked the artificial intelligence boom when it launched its free ChatGPT app in November 2022, and thus built its user base from scratch in under four years.

Every audience metric feeds into the valuation that OpenAI hopes to achieve with an eventual IPO, for which it filed a confidential application in June.

In a sign of tightening competition against Gemini, China's DeepSeek, and its San Francisco rival Anthropic, which makes Claude, ChatGPT's market share for AI assistants fell below 50 percent in the first half of the year, according to research firm Sensor Tower, which uses company data to estimate usage.