The Chinese startup Moonshot has actively entered the competition for leadership in artificial intelligence with the launch of Kimi K3. The company claims that this open-weights model has 2.8 trillion parameters, positioning it as the largest of its kind in the world.
Kimi K3 was developed with a focus on performing tasks that require sophisticated reasoning, long-term programming, and knowledge-intensive activities. According to Moonshot, it represents the first open-weights model to approach the three-trillion parameter mark. In addition to its size, the model features a context window of 1 million tokens, allowing it to process a significantly larger amount of data in a single request compared to previous versions.
The main attributes highlighted by the company include: 2.8 trillion parameters; a 1 million token context window; focus on complex programming and advanced reasoning; and being an open-weights model, which allows for customization by users. This feature enables developers to download, run, and fine-tune the system for various applications, unlike closed models.
Moonshot reported that Kimi K3 demonstrated competitive performance when compared to Fable 5 and surpassed Opus 4.8, GPT 5.6 Sol, and GPT 5.5 in GPU kernel optimization. This optimization is a technique used to improve hardware utilization and reduce latency.
The results were corroborated by external evaluations. The Arena.ai platform ranked the model in the first position in a specific test on web interface creation. Vals AI ranked it second, only behind Fable 5 but ahead of GPT-5.6 Sol. In sequential complex task tests, Artificial Analysis found performance similar to GPT-5.5 and Claude Opus 4.8.
Despite the progress, there are limitations to consider. Lian Jye Su, chief analyst at Omdia, mentioned that the execution cost could be a fraction of the price charged by OpenAI to its clients. However, he stressed that this does not automatically guarantee the best performance. On the other hand, Ryan Fedasiuk, a researcher at the American Enterprise Institute, warned that running a model of this magnitude locally would require equipment estimated to cost hundreds of thousands of dollars.
Reuters reported that Moonshot, supported by Alibaba and Tencent, continues to expand its financial and technological capacity. As reported by Bloomberg, the startup is seeking to raise $2 billion (equivalent to about R$ 11.1 billion), with a valuation close to $30 billion (approximately R$ 166.5 billion) before a potential public offering in Hong Kong. With the launch of Kimi K3, the company strengthens its strategy to compete in the global artificial intelligence landscape during this period of rapid industry evolution.
Choosing the best Xiaomi smartwatch fundamentally depends on the available budget and the type of use intended by the consumer. The brand offers a vast range of products, from low-cost smart bands, ideal for tracking physical activities and monitoring health indicators, to high-end smartwatches in the Xiaomi Watch line, equipped with GPS, AMOLED screens, and advanced features. Meanwhile, the Redmi Watch models are designed to balance price and features, catering to those looking for the best cost-benefit.
Tecnoblog compiled a list of the ten best Xiaomi smartwatches recommended for 2026, detailing the pros, cons, and specific profile of each device.
The Xiaomi Watch 5 is presented as a sophisticated and robust watch, built with a stainless steel case and protected by sapphire glass on both sides of its 1.54-inch AMOLED screen. Its 5 ATM water resistance allows the user to swim in pools or near the coast.
One of the great differentiators of this smartwatch is the Wear OS operating system, which grants access to Play Store applications such as Google Maps, Calendar, and Google Wallet. Additionally, integration with Gemini enables obtaining answers, ideas, or automating tasks. This model has a 930 mAh battery, offering up to 18 days of autonomy in power-saving mode, but requires compatibility with Android smartphones from version 8.0 onwards.
The Redmi Watch 6 features an aluminum alloy case and displays a 2.07-inch AMOLED screen. This display not only provides a wide viewing area but also ensures good readability even under strong sunlight, reaching a maximum brightness of 2,000 nits. As a more basic model, health monitoring is more restricted; however, it offers over 150 training modes and can record essential vital signs, such as heart rate and blood oxygen saturation.
The battery life of the Redmi Watch 6 is notable, potentially lasting up to 24 days of light use.
The Redmi Watch 5 Lite is positioned as a smartwatch focused on the cost-benefit ratio. Despite its plastic casing, it features 5 ATM water resistance and a 1.96-inch AMOLED screen with Always On Display (AOD) function. The watch supports over 150 types of exercise and includes integrated GPS, promoting greater autonomy compared to the phone. Furthermore, it monitors health data such as stress level, heart rate, and blood oxygenation, standing out with a battery that lasts up to 18 days in standard use.
The Redmi Watch 5 Active represents an even more economical alternative compared to the Redmi Watch 5 Lite. Its external finish is plastic, and it uses a 2-inch LCD screen to facilitate interface viewing. This model focuses on basic health monitoring and offers over 140 training modes. The battery promises long duration, reaching up to 18 days, and is compatible with both Android and iOS systems, although it does not have integrated GPS.
The Xiaomi Watch S5 establishes itself as a mid-range smartwatch with high-standard features. It has a stainless steel construction with slim bezels and a 1.48-inch AMOLED screen with a peak brightness of 2,500 nits, facilitating visibility in outdoor environments. An additional strong point is the integrated dual-band GPS, which ensures greater location accuracy. With offline map support, the user will not depend on the smartphone to follow routes during activities such as running, cycling, or skiing.
Other specifications include a battery with up to 21 days of autonomy, heart rate sensor, Wi-Fi support, and 5 ATM water resistance.
The Xiaomi Smart Band 10 Pro, although classified as a smart band, features a 1.74-inch AMOLED screen reminiscent of a smartwatch. Its finish is reinforced with a high-strength aluminum structure and has 5 ATM water protection, allowing for swimming in shallow waters. It is ideal for basic health monitoring, such as heart rate, menstrual cycle, and sleep data. The PPG module has been improved for more accurate heart rate measurements. The band offers good autonomy, reaching 21 days of battery life, in addition to having integrated GPS and over 150 sports modes.
The Xiaomi Smart Band 9 Active is another band suitable for those who need a device to track workouts, basic health data, and phone notifications. It has a lightweight design (only 16.5 g), a 1.47-inch TFT LCD screen, and 5 ATM water resistance, supporting depths of 50 meters. It offers 50 training modes and a battery life of up to 18 days, compatible with Android or iPhone phones.
When selecting the best Xiaomi smartwatch in 2026, it is crucial to weigh factors such as design, durability, functionality, smartphone integration, battery life, as well as price and available budget. It is recommended to prioritize models with resistant materials and high water resistance if engaging in extreme or aquatic sports.
Regarding the screen, one should look for high peak brightness to optimize visibility in the sun, and AMOLED screens are preferred for offering more vibrant colors than LCD displays. For health monitoring, it is suggested to focus on watches with more advanced sensors, unless a smart band meets basic needs. Battery longevity is another important factor, favoring devices that require few recharges. Compatibility with the phone's operating system is vital; for example, iPhone users should not consider the Xiaomi Watch 5 due to incompatibility.
Extra features such as NFC, Wi-Fi, and AI integration should be evaluated based on daily utility. The Xiaomi Watch 5 is highlighted as the best smartwatch with NFC from the brand, using Google Wallet for contactless payments in Brazil. It is also considered the best Xiaomi smartwatch with GPS, equipped with a modern dual-band GNSS chip (L1 + L5), compatible with five satellite systems: BeiDou, GPS, GLONASS, Galileo, and QZSS.
For those seeking an affordable option, the Redmi Watch 5 Active is recommended as an economical smartwatch, offering basic health monitoring and over 140 training modes. Its 2-inch screen provides good visibility despite being LCD technology, and the battery can last up to 18 days.
To confirm if a Xiaomi smartwatch is original, it is necessary to check the quality of the prints and the absence of spelling errors on the packaging, as well as checking for the presence of manuals. It is essential to connect the device to Mi Fitness, Xiaomi's official application; the suggestion of generic third-party apps increases the probability of counterfeiting. Finally, the purchase must be made from reliable stores, and it is recommended to follow TB's findings to find the best prices.
Xiaomi smartwatches have a shape similar to a traditional watch, featuring circular or square screens. These devices go beyond simple health monitoring, offering extensive connectivity with smartphones, integrated GPS, NFC for payments, and AI features. In contrast, Xiaomi smartbands are more compact, resembling bracelets, with oval or rectangular screens, and are focused on basic health monitoring, possessing fewer connectivity features and consequently being cheaper.
The Redmi Watch line encompasses entry-level or mid-range smartwatches, whose focus is cost-benefit. These models generally feature simpler constructions (plastic or aluminum) and more limited health tracking, along with square screens and guaranteed absence of GPS and NFC. The Xiaomi Watch series, on the other hand, consists of premium watches. Besides having round screens, these devices have a more durable build, more features, and more complete connectivity options, making them more comprehensive smartwatches.
Elihle Stali, a young inventor, is transforming the field of assistive technology with his innovative smart glasses for people with visual impairments. Previously, in an interview for TikTok, the 24-year-old Stali appealed to his compatriots in South Africa for help raising funds for a trip to Hong Kong.
His goal was not just to represent his country but to prove that an idea born in East Cape, in the settlement of eMdantsane, could compete with advanced global technologies. Several weeks later, Stali returned home with the future of African assistive technology packed in a briefcase. He established strong business ties with 11 venture investors from India, Brazil, China, the USA, Spain, France, and the UK, and, more importantly, secured a breakthrough international manufacturing partnership.
This alliance will allow for the mass production of Spectacles4TheBlindSA—his revolutionary artificial intelligence-based smart glasses for the blind, moving the technology from a working prototype stage to full-scale commercial production. These glasses are intended to empower millions of people with visual impairments. Stali's mission is inspired by a promise made to his grandmother.
Long before global venture capitalists noticed his company, Stali had a grandson and a grandmother. Growing up in South Africa, Stali watched his grandmother lose independence due to vision problems. Every daily task required assistance: walking to church, navigating the kitchen, or leaving the house. As a teenager, observing the rapid development of artificial intelligence, he asked a key question: what if wearable technology could restore the dignity and independence of millions of people with visual impairments?
This question gave rise to Spectacles4TheBlindSA. The device uses advanced software for object recognition and computer vision to scan the environment in real time. The program then converts visual data into a descriptive audio stream, providing precise navigational cues to the user. Critically important for the African continent is that the software communicates in several local languages, including isiXhosa, Zulu, Sesotho, and Afrikaans, thereby eliminating the huge accessibility barrier left by Western tech giants.
Stali stated, 'I achieved more than I could have imagined. I kept my promise to my grandmother.' Selected as one of 150 leading global innovators, Stali spent three intensive days demonstrating his device to global tech firms. Unlike other creators who relied on presentations and corporate slides, Stali presented his live, functioning prototype directly to global investors. He explained, 'I wanted people to experience it. When visitors stopped at my booth, I didn't rush into a typical sales pitch. I told them about my grandmother. Most people deeply connected with the pure human purpose behind the code.'
Although international manufacturing deals are progressing, specific company names remain under strict confidentiality agreements. For Stali, finding a partner capable of realizing his technical vision was a huge victory. He noted, 'I wanted someone who could truly build this innovation. For me, it was one of the biggest wins. I am only 24 years old. I want to build more than a product. I want to create jobs. I want to create a legacy.'
The trip also highlighted an important business lesson: African ideas require aggressive legal protection. While international commercial negotiations continue under non-disclosure agreements, Stali and his legal team are focused on intellectual property protection. Stali stated plainly, 'We know that historical patterns show that Black inventions are exploited or stolen.' He added that before mass production begins, they are actively registering trademarks and protecting their patent rights.
Paradoxically, while international investors were eager to support the device, Stali found that securing support at home was much harder. Before traveling abroad, he approached several local funding bodies, innovation agencies, and South African government programs. Most of his applications went unanswered. Stali shared, 'I knocked on many doors back home. My dream is not only to help blind people navigate the world safely. It is also to establish a sustainable manufacturing plant here in South Africa to create high-skilled jobs for our youth.'
OpenAI has introduced a new methodology to measure the return that artificial intelligence generates within companies. This proposal aims to replace conventional adoption indicators, such as the number of users or licenses purchased, with an analysis focused on the work that was effectively completed.
The company argues that AI performance should be evaluated by the ratio between the value generated and the cost required to obtain reliable results. The scope of this evaluation is not limited only to the processing cost of the models but covers all effort expended until a task is successfully completed.
According to OpenAI, the technological advancement of AI depends on four fundamental pillars: the volume of useful work executed, the cost associated with each result obtained, the ability to produce reliable answers, and the potential to increase these gains as usage intensifies.
The discussion about the return on AI investments gains a new dimension with the suggestion of a metric called 'useful intelligence per dollar.' This concept aims to determine whether the resources invested in AI are actually providing tangible benefits to professionals and corporations.
OpenAI emphasizes that focusing solely on technical aspects, such as the cost per processed token, does not reflect the total economic impact of the technology. A more accessible model may require more attempts, human revisions, and time to achieve a satisfactory outcome, whereas a more sophisticated solution may complete the same activity in a single execution.
The company maintains that the most relevant indicator is the total cost incurred to complete a task with adequate quality. This calculation must incorporate various elements, such as computational processing, employee contribution, necessary corrections, and rework until the result is ready for use.
The initial evaluation process would involve identifying a specific activity within an organization and clearly defining what constitutes its successful completion. For example, in support teams, the metric could be linked to resolving customer issues; in engineering, it could correspond to code modifications validated in tests; and in legal sectors, it could involve contract review done correctly within the stipulated timeframe.
OpenAI illustrates with the case of a financial team responsible for preparing a projection analysis meeting. In this scenario, AI can assist in locating updated data, organizing information, comparing changes, reviewing spreadsheets, and preparing materials, freeing up specialists to focus more on interpreting findings and making strategic decisions.
Additionally, the company points out that different models meet varied demands. It states that an architecture with multiple options allows for the selection of more agile solutions for simple tasks or the use of more advanced models when complex activities require greater reasoning power.
OpenAI also mentioned the launch of GPT-5.6, which was presented with three modes of use: Sol, characterized as the highest capacity model; Terra, focused on balancing performance and cost; and Luna, aimed at speed and economy. According to the company, the choice of model should be guided by the efficiency in completing the task, and not merely by the unit value of each processing.
Reliability emerges as another crucial aspect in this evaluation. OpenAI indicates that AI implementation tends to progress in stages: initially as a content creation tool, then as a data analysis resource, and subsequently as a system capable of executing phases of processes under human supervision.
To monitor this progression, the company suggests tracking three possible indicators: deliverables ready for use, responses requiring adjustments, and situations where human intervention is necessary to finalize the activity.
More reliable systems contribute to reducing the time spent on verification, corrections, and task repetition. However, expanding AI usage requires prior delineation of boundaries, specifying which data is accessible, which systems can be altered, and when human approval must be obtained.
OpenAI emphasizes that tools intended for the corporate environment must integrate technical competence with robust mechanisms for security, privacy, regulatory compliance, and management. This structure would allow the expansion of artificial intelligence access to vital processes without compromising organizational control.
The last dimension analyzed by the company is the capacity to generate greater value as usage increases. To achieve this, companies must track the quantity of tasks completed with quality, the costs involved, and the average price of each result over time.
In OpenAI's view, advances in infrastructure, more efficient models, improved processing systems, and product enhancements can raise the return obtained per unit of investment. The company establishes this cycle as a continuous sequence: more efficient infrastructure drives research, research generates superior models, models refine products, and products promote greater adoption.
Finally, OpenAI concludes that artificial intelligence should be examined by its ability to expand meaningful work, support decision-making, and free professionals for activities that depend on human judgment, creativity, and experience.
