

iPhone 17 Pro price officially reduced by 1000 yuan

ChatGPT's integration into Siri fell short of expectations; OpenAI reportedly plans legal action against Apple.

With a valuation of $4.65 billion, Tian Yuandong, former director of Meta Fair, has officially launched his startup.

Claude has reinstated access to Lobster, prompting developers to say it's a disguised price increase.

Tencent denies rumors that its top AI executive will leave the company.

The Jensen Huang Foundation donated $108 million in AI computing power.

InStone CEO responds to Luna's high price: 5299 yuan is the US price.

HarmonyOS is in talks to develop Maserati's new energy vehicle.

BMW Brilliance's annual water conservation data released: Reclaimed water replaces tap water, reducing total consumption for three consecutive years.

Intel becomes McLaren's official computing partner

Honda suffered its first annual loss since its listing.

LinkedIn announced layoffs affecting its engineering, product, and marketing departments.

Anthropic CFO: AI has written over 90% of the code; the human role is shifting to "supervisor".

Reports indicate that internal testing of GPT-5.6 has begun, and the flagship model's "ultra-fast mode" will be launched this week.

Honor 600 series scheduled for release on May 25th

On the eve of Google I/O, Gemini 3.2 Flash and several "Agent" models were leaked ahead of time.

Tencent open-sources its AI Agent memory system, reducing token consumption by up to 61%.

Pop Mart denies entering the traditional home appliance market; LABUBU refrigerators reported first-quarter revenue of approximately 6 million yuan.

iPhone 17 Pro price officially reduced by 1000 yuan

Last night, Apple unexpectedly launched a 618 promotional event for the iPhone 17 Pro series on Tmall's "Apple Store Official Flagship Store". The Pro and Pro Max models were reduced by 1,000 yuan, with starting prices of 7,999 yuan and 8,999 yuan respectively.
This event also covers iPad, Apple Watch, AirPods and other series. Non-iPhone products can use surprise coupons ranging from 50 to 100 yuan, which can be combined with 88VIP consumption coupons, store discounts and national subsidies and other multiple benefits.

ChatGPT's integration into Siri fell short of expectations; OpenAI reportedly plans legal action against Apple.

According to Bloomberg, the partnership between Apple and OpenAI has shown clear signs of strain, and OpenAI is working with external law firms to assess possible legal action.
Sources familiar with the matter revealed that OpenAI's legal team is collaborating with external law firms to explore several options, including sending a breach of contract notice to Apple, but no formal lawsuit has been filed yet. OpenAI still hopes to resolve its dispute with Apple out of court, and related legal action is expected to proceed only after Musk's related lawsuits have concluded.
Apple's collaboration with OpenAI began in June 2024, when Apple integrated ChatGPT into Siri and iOS at WWDC, covering scenarios such as text generation, visual intelligence, and the image generation app "TuLeFun".
With a valuation of $4.65 billion, Tian Yuandong, former director of Meta Fair, has officially launched his startup.

Yesterday, Tian Yuandong, former director of Meta FAIR Research Institute, officially announced the co-founding of AI startup Recursive Superintelligence, exactly six months after confirming his departure from Meta.
He posted on X that he rejected offers from several leading AI companies, including OpenAI, xAI, and Anthropic, and chose to start his own business: "While I'm still young, I'm going to be a co-founder of a new startup."
Recursive also announced the completion of a $650 million funding round, valuing the company at $4.65 billion. The round was led by GV (Google Ventures) and Greycroft, with participation from AMD Ventures and Nvidia.
The company's core focus is on "recursive self-improvement," which means building an AI system that can automatically discover knowledge and continuously optimize itself.
Claude has reinstated access to Lobster, prompting developers to say it's a disguised price increase.

Anthropic officially announced yesterday that starting June 15th this year, it will introduce a separate monthly quota for the Agent SDK for Claude subscription users. As a result, third-party Agent tools such as OpenClaw will be allowed to access Claude through subscription accounts again.
According to the new regulations, calls to the Agent SDK will be completely separated from the user's regular subscription limit and replaced by a dedicated monthly quota pool. The quota will be refreshed according to the billing cycle, and any unused portion will not be carried over to the next month.
Pro and Team Standard seats receive a $20 monthly credit; Max 5x and Team Premium seats are $100; Max 20x and Enterprise Premium seats can be up to $200.
This quota is dedicated to Claude Agent SDK projects (Python or TypeScript), claude -p non-interactive commands, Claude Code GitHub Actions integration, and third-party applications certified by the Agent SDK.
This new rule is seen as Anthropic's formal solution to the ban policy implemented on April 4th of this year.
At that time, Boris Cherny, head of Claude Code, announced that Claude Pro and Max subscriptions would no longer cover calls to third-party tools such as OpenClaw, arguing that the usage patterns of such tools were putting unsustainable pressure on Anthropic's computing infrastructure.
The developer community reacted coldly. Some developers bluntly stated that the $20 Agent credit was consumed extremely quickly under high-intensity tasks, with the actual usable amount being far lower than the subscription plan before the ban, and characterized this adjustment as a "disguised price increase."
Tencent denies rumors that its top AI executive will leave the company.

Yesterday, Tencent officially issued a statement refuting rumors circulating online that its "AI leader is about to leave the company." Tencent's PR director, Zhang Jun, stated, "This is outrageous; they've even fabricated all sorts of ridiculous scenarios."
The rumor originated from multiple posts on social media platforms, claiming that the AI chief of a major company was preparing to resign because the company's green business unit had taken away computing resources, and specifically alluding to Yao Shunyu, the head of Tencent's AI division.
The Jensen Huang Foundation donated $108 million in AI computing power.

According to Reuters, the foundation established by Nvidia CEO Jensen Huang and his wife Lori purchased AI computing power from cloud computing company CoreWeave and donated it to several universities and other non-profit research institutions for scientific and artificial intelligence research.
According to a document submitted yesterday, the donated computing power is currently valued at $108.3 million. Nvidia also stated that it plans to provide free engineering services to some of the recipient organizations.
CoreWeave is a cloud computing company focused on AI applications, and the GPUs it provides to its customers are designed and manufactured by Nvidia. In January of this year, Nvidia invested $2 billion in CoreWeave, briefly becoming the company's second-largest shareholder.
InStone CEO responds to Luna's high price: 5299 yuan is the US price.

Yesterday, InStone CEO Liu Jingkang responded to the pricing issue of its flagship dual-camera gimbal camera, Luna Ultra, stating that the rumored price of 5299 yuan for the single unit is true, "but that's the price in the United States. I wouldn't dare to sell it at that price in China, no matter how bold I am."
Previously, the blogger "Digital Chat Station" revealed that the machine was expected to be priced at 5,299 yuan and the all-in-one package was expected to be priced at 6,499 yuan, which immediately sparked doubts online about the "high price".
HarmonyOS is in talks to develop Maserati's new energy vehicle.

According to CloudInsight, Huawei HarmonyOS, JAC Motors, Stellantis Group and its Maserati brand are in talks to jointly develop new energy vehicles under the Maserati brand.
According to the plan, the four-party cooperation model is highly similar to the "Five Realms" of HarmonyOS: Huawei leads the product definition and provides core technologies, JAC jointly develops and is responsible for production and manufacturing, and Maserati provides styling design and brand endorsement.
The collaborative models will be divided into domestic and overseas versions. The domestic version will belong to the Zunjie brand, while the overseas version will be branded as Maserati. The first model is currently in the styling design stage and is planned to be mass-produced in the second half of next year.
BMW Brilliance's annual water conservation data released: Reclaimed water replaces tap water, reducing total consumption for three consecutive years.

BMW Brilliance released its annual water conservation data yesterday, showing that the water consumption per unit at its Shenyang production base will be reduced to 1.4 cubic meters per unit in 2025, a year-on-year decrease of 10.8%, marking the eighth consecutive year of steady decline.
In terms of total water consumption, the total water intake of BMW Brilliance's Shenyang plant decreased to 752,000 cubic meters throughout the year, a reduction of over 40% compared to 2023, marking the third consecutive year of decline. The use of reclaimed water exceeded 680,000 cubic meters, an increase of 71.9% year-on-year. Officials stated that the water saved could meet the annual domestic water needs of nearly 9,300 households in Shenyang.
BMW Brilliance's Dadong Plant North Plant has achieved zero tap water consumption in all production processes, replacing it entirely with reclaimed water; the cooling towers in the painting workshops of the Tiexi Plant and Lida Plant have also expanded to use municipal reclaimed water, and the irrigation of the plant's green areas had already achieved 100% use of municipal reclaimed water by 2024.
Intel becomes McLaren's official computing partner

Intel announced yesterday a multi-year strategic partnership with the McLaren Racing team, officially becoming the official computing partner of the McLaren MasterCard Formula 1 team, Arrow McLaren Indy team, and McLaren F1 Sim Team.
Under the cooperation agreement, Intel Xeon processors and Core Ultra processors will provide computing power support for McLaren's core performance workloads, covering areas such as computational fluid dynamics, aerodynamic analysis, vehicle dynamics simulation, and race strategy analysis.
Intel's technology will also be applied to scenarios such as edge computing, digital twins, and AI predictive modeling.
Honda suffered its first annual loss since its listing.

Yesterday, Honda released its consolidated financial results for the fiscal year ending March 2026. Revenue for the period was 21,796.6 billion yen, a slight increase of 0.5% year-on-year; net loss attributable to shareholders was 423.9 billion yen , compared to a net profit attributable to shareholders of 835.8 billion yen in the same period of the previous fiscal year, turning from profit to loss.
The main reason for the loss was a significant adjustment to its electrification strategy. During the fiscal year, Honda decided to cancel the development and launch plans for several electric vehicle models in North America, resulting in impairment losses, disposal losses, and contract compensation totaling approximately ¥1.5777 trillion, all of which were recorded in the automotive business. The motorcycle business was the only segment to achieve positive profitability this fiscal year.
Looking ahead to next year, the company expects revenue of 23.15 trillion yen and net profit attributable to shareholders to recover to 260 billion yen.
LinkedIn announced layoffs affecting its engineering, product, and marketing departments.
According to Bloomberg, LinkedIn, the professional networking platform, announced layoffs yesterday, affecting multiple functional departments including engineering, product, and marketing.
In a memo to employees, LinkedIn CEO Daniel Shapero stated that the company needs to provide greater value to users and achieve higher profitability, and this organizational restructuring aims to position the company for future development.
A LinkedIn spokesperson later confirmed the changes, stating that the adjustments were part of "routine business planning," but did not disclose the specific number of layoffs.
Anthropic CFO: AI has written over 90% of the code; the human role is shifting to "supervisor".

According to a Business Insider report yesterday, Anthropic CFO Krishna Rao revealed on the podcast "Invest Like the Best" that Claude currently handles over 90% of the company's coding work and is deeply involved in financial processes, driving a shift in white-collar work from "execution" to "supervision".
Rao also revealed that the company has been using Claude to generate financial statements, and the monthly financial review process is now 90% to 95% complete before manual intervention. He mentioned that the generation time for some internal reports has been reduced from several hours to 30 minutes.
Regarding staffing, Rao described the introduction of AI as a productivity "accelerator." He stated that because of AI's involvement, employees were able to shift more of their energy from information gathering to decision-making, which actually led Anthropic to expand its hiring scale.
We hired more people because there's no job that can't be done.
Rao also described a new work model: employees are increasingly taking on the role of "managers" of AI systems, with teams deploying multiple batches of AI agents to process projects in parallel. He summarized this trend as "everyone becoming a manager to some extent," and believes that the resulting productivity potential "is only just beginning to emerge."
He also emphasized that Anthropic values "talent density" rather than "talent scale," believing that a high density of top AI research talent combined with the strongest models is the truly competitive combination.

Reports indicate that internal testing of GPT-5.6 has begun, and the flagship model's "ultra-fast mode" will be launched this week.

Renowned leaker Leo revealed yesterday that GPT-5.6 development is in full swing and is expected to be officially released as early as next month. Developers discovered two internal codenames—ember-alpha and beacon-alpha—as well as a routing mapping record in OpenAI's internal Codex logs.
The logs show that the vast majority of model calls still point to GPT-5.5, but one entry explicitly points to GPT-5.6, suggesting that the Codex environment may be testing a new model.
In addition, the whistleblower Chetaslua claimed that OpenAI plans to launch a "super-fast mode" for its flagship model this Thursday, which could improve response speed by 2 to 3 times, primarily targeting latency-sensitive tasks.
Honor 600 series scheduled for release on May 25th

Honor announced yesterday that its Honor 600 series will be officially launched on May 25th at 19:00.
Official teaser information shows that the Honor 600 series focuses on "4K flash mirrorless Live" imaging, equipped with a 200-megapixel ultra-clear large-sensor main camera, and built-in 8600mAh Qinghai Lake large battery.
On the eve of Google I/O, Gemini 3.2 Flash and several "Agent" models were leaked ahead of time.

According to Sources, Google plans to officially release its next-generation lightweight model, Gemini 3.2 Flash, at the Google I/O conference on May 20. Its overall performance is roughly equivalent to that of GPT-5.5.
Abacus.AI CEO Bindu Reddy revealed that the Gemini 3.2 Flash reportedly achieves approximately 92% of the performance of GPT-5.5 in coding and inference tasks, while its inference cost is only one-fifteenth to one-twentieth of the latter, with most query latency below 200 milliseconds.
According to leaker Leo (@synthwavedd), Google DeepMind apparently accidentally pushed a batch of unreleased models to its production API, including:
The Gemini 3 Flash, Gemini 3 Pro Image, Gemini 3.1 Flash Image, Gemini 3.1 Flash Lite, Gemini 3.1 Flash TTS, Gemini 3.1 Pro, and the music generation models Lyria 3 Clip and Lyria 3 Pro, as well as a video generation model codenamed "Omni".
It is worth noting that all of the above models have variant versions with the suffix "agent".
Tencent open-sources its AI Agent memory system, reducing token consumption by up to 61%.

Tencent Cloud's database team yesterday officially open-sourced TencentDB Agent Memory, a local-first memory engine for AI agents, supporting integration with OpenClaw and Hermes Agent. The core design of this system lies in splitting the memory into two independent structures:
- The long-term memory portion is divided into L0 original dialogue, L1 atomic facts, L2 scene blocks, and L3 user profile, which are then progressively accumulated layer by layer.
- Short-term task memory externalizes the lengthy tool log to a refs file, writes a summary of steps to jsonl, and preserves the task structure and node indexes using a Mermaid canvas.
Official benchmark tests show that after integrating OpenClaw, the token consumption for WideSearch tasks decreased from 221.31M to 85.64M, a reduction of 61.38%, while the task pass rate increased by 51.52%. On the long-term memory evaluation set PersonaMem, the accuracy improved from 48% to 76%.
GitHub: github.com/Tencent/TencentDB-Agent-Memory

lululemon launches lightweight sun protection jacket series

lululemon recently released a lightweight sun-protective jacket series, featuring UPF 40+ sun protection, which can block more than 97.5% of ultraviolet rays, while also being lightweight and breathable.
- For women, the self-packing sun protection jacket is made of Glyde fabric, which has triple performance of UV protection, rainproof and windproof. The Goal Smasher sports sun protection jacket is designed for running, with elastic splicing structure and partial opening design, emphasizing a cool experience when running flexibly.
- For men's styles, Stash and Dash
The self-packing sun protection jacket features multiple pockets and can also be packed away as a portable pouch; another lightweight version of the sun protection jacket also uses Glyde fabric, and the back ventilation design provides stable protection for changeable weather.
Pop Mart denies entering the traditional home appliance market; LABUBU refrigerators reported first-quarter revenue of approximately 6 million yuan.

Recently, Pop Mart's Chief Operating Officer, Si De, stated in the Q1 2026 business call that the company has no plans to expand into traditional home appliances. In Q1 of this year, the sales revenue of LABUBU refrigerators was approximately RMB 6 million, accounting for a very small percentage of the total revenue.
According to Si De, Pop Mart's home appliance business is currently in a very early stage. In the future, it will combine small home appliances with lifestyle, IP and trendy toy culture to create an important supplementary category.
Sidde stated that the company will not accelerate the development of its home appliance business, but hopes to refine its products and supply chain in a more stable way, "taking more time and patience to continuously improve this category."
CASETiFY collaborates with a Japanese illustrator to launch the "nokonoko uchinoko" series.

CASETiFY announced yesterday that it has launched a new collaboration series, "nokonoko uchinoko," with Japanese illustrator Saori Iijima. The series is inspired by the imagery of soft, baked bread and uses gentle brushstrokes to depict everyday emotions.
Saori Iijima is an independent illustrator from Japan, known for her delicate and soft art style. She has previously collaborated with several brands. CASETiFY regularly collaborates with global artists and IPs, and this artist series is one of their regular collaborative projects.

Netflix's "East of Eden" trailer released, set to premiere this fall.

According to Douban Movie, Netflix released a Chinese-subtitled trailer for the series "East of Eden" yesterday, and the official announcement confirmed that the series will be released this fall, consisting of 7 episodes.
The series is adapted from John Steinbeck’s classic novel of the same name published in 1952. Set in the Salinas River Valley of California, it tells the story of the intertwined fates of the Trask and Hamilton families across generations, spanning from the American Civil War to the end of World War I.
The Big Bang Theory spin-off series "Stuart Failed to Save the Universe" has released its release date and trailer.

The trailer for "Stuart Failed to Save the Universe," a spin-off of "The Big Bang Theory," has been released, announcing its release date.
The story centers on Stuart Bloom, the owner of a comic book store.
He accidentally damages a device built by Sheldon and Leonard, triggering a multiverse-level apocalyptic crisis. He is then forced to work with his girlfriend Denise, geologist Bert, and quantum physicist Barry Kripke to repair reality, encountering characters from various parallel universe versions of The Big Bang along the way.
K-POP: Witcher Girls Global Tour Officially Announced

Netflix and AEG Presents recently announced that they will adapt the animated film "K-POP: The Witcher" into a world tour, stating that they will bring elements of the film to the stage in a "dynamic and memorable way".
K-POP: The Witcher won two awards at this year's Academy Awards: Best Animated Feature and Best Original Song. The winning track "Golden" became the first K-pop song in history to win the Academy Award for Best Original Song.
The film's soundtrack also won a Grammy Award and topped the Billboard Hot 100 chart, becoming the best-selling soundtrack worldwide last year.
#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.