i don’t know how they’ve done it but China’s basically been able to replicate every major American AI model within 3 months of official release and it’s ALWAYS cheaper and smaller (efficient) It isn’t just LLMs either, e.g. tencent just dropped 2 models, a world simulator and live foreign language translator that goes head to head with the models Google dropped last week lmao (and they’re 7B vs whatever gargantuan behemoth Google built) my point is, the stereotype I had of China when it came to competing with America on tech has been completely shattered. I thought maybe DeepSeek was luck, thought the same when Alibaba dropped QWEN but now they’ve had too many dubs to deny, they’ve got amazing talent and we should be 100% dialled in and not dismissing them because it’s “Chinese made” TikTok was the warning shot (generated more rev on the social media ad side than meta last few months) and I’m guessing we’ll continue to see outperformance in AI going forwards. final reminder: over 50% of the worlds best AI talent resides in China. These individuals are worth their weight in gold and are being churned out from Tsinghua university. Chinas also commands a huge power reserve (data centers) for training models vs USA. Once they crack the chip manufacturing it’s game over. We have until then to catch up.

According to BiJie Wang news, on August 15, Alitong Yiqianwen announced the contextual expansion of 1 million tokens for Qwen3, the upgrade of Qwen Chat Deep Research, and the release of Qwen-Image refined on ComfyUI. In addition, the Qwen-Image image editing model is set to launch soon.

The local large model was tinkered with all night, and the final choice was made: 1. Main model: Qwen-30B-Instruct, it's sufficient for daily tasks. (Instruction following is very good) 2. Inference backup: Left an 8-bit modified version of GPT OSS mlx. The 4-bit version of GPT-OSS doesn't work well, and maximizing the inference budget across three tiers is not very meaningful. 3. Coder in all directions, planning to use SOTA flagship directly without considering local models (after all, it's work)

According to news from BiJie, the decentralized AI training platform FLock.io has jointly initiated the Qwen x FLock x Base AI hackathon with Alibaba's open source large language model team Qwen and the blockchain infrastructure platform Base. This hackathon brings together developer clubs from Korea's "Ivy League" SKY and KAIST, promoting innovation and practical application of AI models through federated learning technology.

Everyone’s talking about onchain AI. But most of it’s just hype. It replaced Claude as the default model on Base months ago. Now it's officially listed in the Base Builder Hub, making it the only AI infra directly integrated into Coinbase’s L2. But it goes deeper 👇 ▪️ Backed by the Ethereum Foundation ▪️ Partnered with Alibaba’s Qwen model ▪️ Running live across Base’s onchain AI system ▪️ Federated learning with real privacy It’s building infra that actually works onchain, in real-time.

According to CoinWorld, the latest research paper proposed by the decentralized AI training platform FLock.io has been successfully selected for the International Conference on Machine Learning (ICML) 2025 CFAgentic Workshop. This achievement stems from the long-term accumulation of the FLock team in their product development and has been gradually realized through the gmFLOCK task incentive model, as well as collaborations with Alibaba's Tongyi Qwen model, Base MC, and other ecosystem partners to implement on-chain infrastructure. ICML, NeurIPS, and ICLR are the three top conferences in the AI field, and the selection criteria are extremely strict. [CoinWorld]

Hello, brothers! Today I'm here to tell you about Hyperbolic Labs—the perfect collision of AI and Blockchain, the future is promising! This project has been quite popular on Twitter (@hyperbolic_labs), combining AI and Blockchain, aiming to make AI Computing Power as accessible as OCN, available for rent at any time, and at a very cheap price! Let me explain this project to you clearly, and also discuss why it might be a potential stock worth following! #COOKIE #CryptoNews @KaitoAI @hyperbolic_labs @cookiedotfun 1. What does Hyperbolic Labs do? In simple terms, Hyperbolic Labs wants to do one thing: make AI computing power as simple and cheap as ordering takeout! Right now, AI is incredibly popular, but the GPU computing power needed to train large models and run inference is ridiculously expensive, making it unaffordable for ordinary developers and researchers. Hyperbolic Labs aims to solve this problem by creating a decentralized AI cloud platform that integrates global GPU resources through Blockchain technology. Anyone with idle GPUs can rent them out, and anyone in need of computing power can rent them anytime, at prices up to 75%-80% lower than traditional cloud services (like AWS, Google Cloud)! Their slogan is: "Making AI accessible and empowering global innovators!" This may not be just empty talk; they have secured collaborations with big names like Hugging Face, Vercel, Google, Quora, Stanford, and UC Berkeley, indicating a significant level of market recognition. 2. Core Highlights: Why is Hyperbolic Labs so awesome? Ultra-low cost GPU rental Hyperbolic Labs' platform allows you to rent top-tier GPUs like NVIDIA's A100 and H100 for just $0.16 per hour, with a startup time of less than 1 minute! It's 75% cheaper than traditional cloud services, and it supports pay-as-you-go, so you can leave when you're done without signing a long-term contract. Want to run a large model inference? Just click a button and you're good to go! Decentralization + Blockchain, safe and transparent The project uses Blockchain technology to manage global GPU resources, ensuring that the allocation and use of Computing Power are open and transparent. They also emphasize data privacy and result verification, ensuring that your data will not be misused and the results produced are reliable, through encryption technology and decentralized nodes. This is a significant pain point in the AI field, and they have managed to capture the core issue. Compatible with mainstream AI ecosystems The API of Hyperbolic is highly compatible with OpenAI and supports the latest open-source large models, such as DeepSeek, Qwen, Llama 3.2, etc. Whether running text, image, or multimodal models, seamless integration is achievable. This is truly a boon for developers, saving them the hassle of re-adaptation. Community-driven + Open Source Spirit Hyperbolic Labs often mentions their open-source projects on Twitter, such as Hyperbolic-AgentKit (an AI writing and research tool) and Hyper-dOS (a distributed operating system). These tools are all open-source and have attracted over 200,000 developers to join the community, with the ecosystem already taking shape. Top teams and endorsements The team is full of hardcore experts: there are AI professors from UC Berkeley, co-founders of Databricks, and engineers who have worked on decentralized solutions and GPU optimization at Ava Labs and Crusoe. The investors are also strong, with Polychain Capital and Lightspeed Faction leading a $7 million seed round, which is no small amount in the AI + Blockchain space. 3. The Market Potential of Hyperbolic Labs AI is now the top trend in global technology, with Computing Power demand doubling every year, but GPU resources are concentrated in the hands of a few large companies, making it impossible for ordinary players to get access. Hyperbolic Labs utilizes idle GPUs through a decentralized platform, essentially creating a "GPU sharing economy." This not only significantly reduces costs but also allows more small and medium-sized teams, independent developers, and even academic institutions to access top-notch Computing Power, promoting the widespread innovation of AI. They mentioned on Twitter that the platform has already supported popular AI applications like Chatbot Arena, which runs on open-source models and delivers results comparable to those of large companies but at a much lower cost. This model is highly attractive for startups and researchers. Speaking of Blockchain, Hyperbolic manages Computing Power and incentive mechanisms in a decentralized way, allowing participants (those providing GPUs) to earn profits through the platform, creating a win-win ecosystem. Their incentive model also incorporates the design of crypto-economics, and there may be a token mechanism in the future (although it hasn't been officially announced yet), which is a potential explosive point for cryptocurrency enthusiasts. 4. Risks and Challenges Of course, the project is not without flaws, and the veteran in the crypto world has to remind you of a few points: Intense competition: The AI cloud market has major players like AWS, Google Cloud, and Azure, as well as blockchain + Computing Power competitors like Akash Network and Render Network. For Hyperbolic to break through, it must rely on technical and community strength. Regulatory Uncertainty: Both AI and Blockchain are regulatory-sensitive areas, and global policy changes may impact projects. Token Uncertain: Currently, Hyperbolic has not disclosed its token plans, and cryptocurrency players may need to wait and see how they design their economic model. 5. Investment and Participation Opportunities To developers: If you are an AI developer, Hyperbolic's low-cost Computing Power and open-source tools are definitely worth a try. Visit their official website ( To cryptocurrency players: Although there are no tokens at the moment, from the tweets and Medium articles, Hyperbolic is actively expanding its community and is likely to introduce incentive mechanisms in the future, possibly through airdrops or node rewards. It is recommended to follow their Discord and Twitter to seize the opportunity as soon as possible. To GPU holders: Brothers with idle GPUs can consider joining their network to rent them out for some extra cash. The platform supports flexible payment methods, including credit cards and cryptocurrencies. 6. Conclusion: Is Hyperbolic Labs worth following? Hyperbolic Labs is definitely a dark horse in the AI + Blockchain track! They have identified the pain point of expensive and hard-to-obtain AI Computing Power, and provided a solution in a decentralized manner. Backed by top teams and investment institutions, the community is highly active, and the technology is being implemented quickly (over 200,000 developers are already using it). Although there are no tokens yet, based on the dynamics on Twitter, there will definitely be big moves in the future. Cryptocurrency players can wait and follow their community and official website updates. Overall, this project has solid technology and the imagination space of the cryptocurrency world. In the short term, it can save money for AI, and in the long term, it may be an ecological opportunity. Brothers, those who are interested should quickly check out @hyperbolic_labs's Twitter and join the Discord to chat. Opportunities in the crypto world never wait for anyone!

Why do we need multiple AIs at the same time? - Using a single AI is prone to bias or errors. – Mira uses a lot of AI to compare results → eliminates misinformation (hallucination) → Helps increase accuracy, safer thanks to "collective intelligence" Which AIs are participating? Chatgpt (OpenAI): strong in complex reasoning DeepSeek: good at math and logic Meta (Llama): reflects multiple perspectives Alibaba Qwen: multilingual support Anthropic (Claude): ethical priority, safety level Sentient: Decentralized AI, operated by the community Nous Research: AI specialized in each field 🤖 How Mira uses this model: Mira operates a private verification network where multiple AIs validate each other. For example: when AI performs automated transactions, other AIs will verify whether the order is accurate to prevent abnormal operations => Enhance the reliability of all onchain actions. The core is Verifiable Automation – automated but always verifiable. ✅ In summary: Mira creates a transparent verification system by utilizing the collective intelligence of multiple AIs → ensuring that every action taken by the AIs is reliable and responsible.

Here is ur Answer: For this to be accomplished, diverse and reliable AI models are use to power the consensus mechanism. They leverage multiple AI models( OpenAI, DeepSeek, Meta AI, Alibaba Qwen, Anthropic, Sentient & Nous Research) They use their training orientations to cross-verify outputs and eliminate hallucinations. This improves verification accuracy and removes single points of failure This collective intelligence enables truly trustless AI verification at scale. GMIRA

According to a report by Jin10 on May 12, Deep Tide TechFlow news states that the Ali Tongyi Qwen large model has become the foundation for AI development in Japan, as mentioned in a recent public report by the Nikkei. In the comprehensive ranking of AI models evaluated by the Nikkei, Tongyi Qwen Qwen2.5-Max ranked sixth, surpassing well-known large models such as DeepSeek-V3 and OpenAI o3-mini from both domestic and international sources. The media mentioned that many emerging enterprises in Japan are developing enterprise large models based on Qwen. So far, Ali Tongyi has open-sourced over 200 models, with global downloads exceeding 300 million times, and the number of derivative models of Qwen has surpassed 100,000, making it the world's first open-source model.

The Nikkei recently reported that Alibaba's Tongyi Qwen large model has become the foundation of AI development in Japan. In the comprehensive ranking of AI models evaluated by Nikkei, Tongyi Qwen Qwen2.5-Max ranked sixth, surpassing well-known large models like DeepSeek-V3 and OpenAI o3-mini from both domestic and international sources. The media mentioned that a large number of emerging enterprises in Japan are developing enterprise large models based on Qwen. So far, Alibaba's Tongyi has open-sourced over 200 models, with global downloads exceeding 300 million times, and the number of derivative models of Qwen has surpassed 100,000, making it the world's first open-source model.

Xiaomi releases its first inference large model MiMo-7B and announces full Open Source. According to the news of Haschain, Xiaomi announced the launch of its first large model designed for inference, MiMo-7B, which is fully open-sourced. The model outperformed OpenAI o1-mini and Ali Qwen-32B in AIME mathematical reasoning and LiveCodeBench programming evaluations. MiMo uses large-scale inference data pre-training combined with reinforcement learning post-training to improve inference ability, and innovatively introduces the Test Difficulty Driven Reward algorithm, which shows the potential of reinforcement learning in similar 7B models. The series has been open-sourced to Hugging Face, and related technical reports have been released at the same time. #HuggingFace #Open Source #AI #MachineLearning #ReinforcementLearning

Alibaba Open Source Next Generation Tongyi Qwen Model Qwen3 According to news from HashiChain and reports from People's Finance, on the early morning of April 29, Alibaba open sourced the next-generation Tongyi Qianwen model Qwen3 (referred to as Qianwen3), with a parameter count only 1/3 of DeepSeek-R1, significantly reducing costs and outperforming models like OpenAI-o1. #Blockchain #Cryptocurrency #OpenSource #Qwen3 #Alibaba

$BABA: Strong Qwen 3 AI Announcement Sentiment: Very Positive Alibaba unveiled its “Qwen 3” AI model family, which reportedly outperforms top-tier competitors. This move showcases its R&D strength and positions the company for broader AI adoption.

🖤 Officially partnered with Alibaba's large model Qwen (k Wen) for a strategic cooperation — this is the first time a Web3 AI project has truly penetrated the top AI circle of Web2. In summary: $FLock is no longer just a side project on the blockchain; it has entered the mainstream AI arena and has become the first project to break down the barriers between circles. To be honest, many people used to think about Web3 AI like this: "Sounds pretty cool, but what is it actually useful for?" Now that's great, Alibaba is directly endorsing it, and they will also work on model training, data privacy protection, and even promote industry standards. Why FLock? Moreover, it's not just talk; they are really running demos, going on-chain, and training models. Now Alibaba is also getting involved with it, which is highly significant. There has been too much Web3 narrative, this time it's not just empty promises, it's serious. Can FLock still seize this opportunity in the future? I don't know, but this has indeed opened up a new landscape.

⚡ Officially announced a strategic partnership with Alibaba's AI large model Qwen (k Qwen). This collaboration signifies that the Web3 decentralized AI training mechanism has been adopted by mainstream Web2 AI companies for the first time, marking the transition from proof of concept to actual industrial application.

Techub News reports that, according to The Block, the decentralized AI model training platform FLock.io has reached a partnership with Alibaba Cloud. FLock.io will utilize Alibaba Cloud's large language model (LLM) Qwen and cloud technology to create advanced AI models for specific domains and general purposes. In the future, both parties will leverage their respective technological expertise to explore research opportunities and use cases in the field of artificial intelligence. In addition to technological integration, FLock.io will also collaborate with Alibaba Cloud to promote joint research programs, leveraging collective knowledge and resources to explore new applications of centralized and decentralized artificial intelligence.

CoinVoice has recently learned that, according to Coingecko market data, FLOCK has risen 60.5% in the last 24 hours, possibly influenced by the announcement of a strategic partnership with Qwen, a leading open source large language model series under Alibaba Cloud.

CoinVoice has recently learned that, according to Coingecko market data, FLOCK has risen 60.5% in the last 24 hours, possibly influenced by the announcement of a strategic partnership with Qwen, a leading open source large language model series under Alibaba Cloud.

#BOB2# #Qwen# #PI# HOW IS THE MARKET TODAY?

#QWEN# when moon..

#ARB#AI today discovered 1. The Federal Reserve's tapering slowdown has triggered a rebound in Crypto Assets, with Bitcoin breaking through $88,000 2. Arrum DAO is reviewing a $225 million ARB gaming incentive program, questioning mismanagement 3. Qwen released a 32B multimodal large model, outperforming the 72B version 4. Movement Foundation: $38 million will be used to repurchase MOVE Tokens 5. The "AI Idol Debut" competition will be revealed, with AI agents expected to have real connections with humans.

AI today found 1. The Federal Reserve's tapering slowdown triggers a rebound in Crypto Assets, with Bitcoin breaking through $88,000 2. Arrum DAO is reviewing the withdrawal of $225 million ARB gaming incentive plan, questioning mismanagement 3. Qwen releases 32B multimodal large model, performance surpassing the 72B version 4. Movement Foundation: $38 million will be used to repurchase MOVE Token 5. The "AI Idol Debut" competition will be revealed, AI agents are expected to have a real connection with humans.

AI today found 1. The Federal Reserve's tapering slowdown triggers a rebound in Crypto Assets, with Bitcoin breaking through $88,000 2. Arrum DAO is reviewing the withdrawal of $225 million ARB gaming incentive plan, questioning mismanagement 3. Qwen releases 32B multimodal large model, performance surpassing the 72B version 4. Movement Foundation: $38 million will be used to repurchase MOVE Token 5. The "AI Idol Debut" competition will be revealed, AI agents are expected to have a real connection with humans.

Good job, Aether! Other projects are just eyeing the Web3 market ( suckers ), but Aether is targeting the domestic AI market. As an AI computing power infrastructure, Aether does have some advantages for domestic AI. First, in terms of performance, Aethir's computing power currently covers 95 countries, with over 400,000 enterprise-level GPUs. Aether's edge computing technology helps Chinese AI products achieve high-performance decentralized computing power, addressing the domestic shortage of computing power. First, in terms of cost-effectiveness, the Aether decentralized computing network breaks the monopoly profits, which is beneficial for lowering the price of computing power. As a more cost-effective computing service, it has certain compatibility with small and medium-sized AI enterprises or teams in China. Third, in terms of resources (, they are called resources in the Web2 field, and can also be called ecosystems in the Web3 field ). Aether has always provided enterprise-level computing power aimed at the Web2 market, establishing long-term strategic partnerships with clients in fields such as gaming and AI. Aether officially provided three examples: Manus AI, Qwen, and ERNIE. These AI models or products are suitable for training and inference of AI models using Aether. I hope more and more domestic AI companies or teams become clients of Aether. On one hand, I hope to see domestic AI reach new heights. On the other hand, I hope Aether can gain more empowerment from the Web2 field and further break the curse of the zero-sum game in Web3!

China's AI Revolution: Aethir Helps AI To da moon China's AI field is no longer "chasing others", but is beginning to "set its own rules". Homegrown projects like Manus AI, Alibaba's Qwen and Baidu's ERNIE have already demonstrated their technological prowess. These AIs are no longer "little assistants" who can only answer questions, but "all-rounders" who can develop solutions across platforms and automate complex tasks. AI is moving from a "tool" to an "all-rounder". However, the rapid development of AI is inseparable from the support of powerful computing power, while the traditional cloud computing model has high cost, high latency, and limited scalability. At this time, Aethir, as a globally distributed GPU cloud infrastructure, stood up and provided high-performance, low-cost computing solutions for the AI industry, becoming the "driving force" behind the development of AI. The power of Aethir: high-performance, low-cost distributed computing. Aethir distributes computing resources across edge nodes globally through a decentralized GPU network, significantly reducing latency and costs. For autonomous agent systems like Manus AI, Aethir's high-performance computing power can accelerate task execution, making the system more stable; for models like Qwen and ERNIE, Aethir's distributed architecture can also support their multimodal and inference capabilities to scale globally. More importantly, Aethir's low-cost computing power allows small and medium-sized AI enterprises to easily enter the market without worrying about GPU costs being too high, which has become a "roadblock." Aethir: A bridge connecting China and the world Aethir is not only a technological "enabler" but also a "bridge" between the Chinese AI industry and the global market. Its global edge deployment capability can help Chinese AI projects enter overseas markets. For example, through the "Belt and Road" initiative, it promotes AI technology to developing countries for use in agriculture, logistics, and other fields. At the same time, Aethir's decentralized model avoids the monopolization of computing power resources by a few giants, ensuring fair competition and healthy development in the AI industry. Ecological cooperation and global potential The collaboration between Aethir and projects like Manus AI, Qwen, and ERNIE is driving Chinese AI technology onto the global stage. By providing efficient and low-cost computing power support, Aethir not only enhances the international competitiveness of China's AI industry but also brings new vitality to the global AI sector. Personal Summary I personally believe that China's AI revolution is changing the global AI landscape, and Aethir, as a leader in distributed GPU cloud infrastructure, is providing crucial support for this transformation. By empowering AI development, reducing computing costs, and connecting global markets, Aethir is driving the rise of Chinese AI technology and opening new doors for innovation and proliferation in the global AI industry. In the future, in-depth collaboration between Aethir and more AI projects will further expand the global influence of Chinese AI technology.

China's AI Evolution: From Catching Up to Defining, How Aethir Propels the Comprehensive Explosion of the AI Industry? China's AI industry is undergoing a paradigm shift, no longer simply "catching up to OpenAI," but directly setting global standards. From Manus AI's fully autonomous agents to Alibaba Qwen 2.5-Max surpassing GPT-40, and Baidu ERNIE 4.5 & X1 low-cost high-performance models, these AI giants are changing the global game. AI Evolution: From "Tool" to "Autonomous Agent" Do you remember the era when AI was just an auxiliary tool? Manus AI has completely changed this perception. It not only helps you plan your trips but also automatically books flights, hotels, and even generates personalized itineraries. Even more surprisingly, it can automatically build websites, analyze stocks, and manage assets, completely without human intervention. This is an important leap for AI from "executing commands" to "autonomous decision-making." 🔹 Manus AI —— The world's first autonomous intelligent agent, integrating multiple models such as Qwen, Claude, etc., to enable AI to automatically perform tasks. 🔹 Qwen 2.5-Max —— The leader of Alibaba's LLM, surpassing GPT-40, open source, promoting AI equality. Computing Power Bottleneck: Where are the Accelerators for AI? The explosion of AI in China has showcased a leap in technological strength to the world, but an issue that cannot be ignored is emerging - the cost of computing power. Training LLMs requires vast GPU resources, and traditional cloud computing models are expensive and often constrained by hardware supply chains. In response to this challenge, Aethir offers a decentralized GPU cloud architecture that provides low-cost, highly elastic computing power, which could become a core pillar of the AI industry chain. 🔹 Decentralized architecture - Distributed GPU network that reduces computing costs while ensuring stability. 🔹 Real-time dynamic provisioning: Provide AI computing resources on demand and adapt to large models such as Manus, Qwen, and ERNIE 🔹 Global Edge Computing - AI is not limited to local applications, but can reach the global market and empower overseas expansion. AI + Aethir: A New Paradigm of Industrial Collaboration If Manus is the "AI productivity tool", Qwen is the "language understanding core", and ERNIE is the "efficient reasoning engine", then Aethir may be their computational foundation. Aethir's decentralized GPU cloud computing not only helps the Chinese AI industry break through GPU supply chain limitations but also promotes the outbound development of AI technology: 🔹 Helping AI companies reduce GPU costs - avoiding the "Matthew effect" caused by computational power thresholds. 🔹 Globalization empowers AI implementation - from Southeast Asian logistics to African smart agriculture, the influence of AI is expanding. 🔹 Flexible adaptation to AI computing needs — Enterprise-level computing power support + On-demand allocation for startups, driving AI innovation. Conclusion: The key role played by Aethir in the AI revolution. The rise of the AI industry in China is beyond doubt, but the next question is: who can support it to run further? AI requires powerful computing power, and Aethir may be the most underestimated disruptor in this wave of change. As Manus, Qwen, and ERNIE continue to push the boundaries of AI capabilities, Aethir's globally distributed GPU cloud might just be the "hidden driver" that ensures this AI race continues to accelerate. China's AI is redefining global standards, and Aethir may be the key piece that makes it all possible.

Aethir: The AI-powered Computing Power engine connecting China with global innovation. 🔔 Introduction: Today's AI is no longer just a simple "question-answering" tool, but a multi-agent system capable of formulating plans and automatically executing tasks across platforms. China's AI is no longer a follower but is beginning to set its own standards. Local projects like Manus AI, Qwen, and ERNIE are demonstrating their strength and have already started sprinting in their respective tracks. 1️⃣ Computing Power cost, the biggest barrier to AI entrepreneurship But no matter how strong AI is, without Computing Power it is useless. GPU prices are ridiculously high, becoming a "roadblock" for AI startups, especially in training large models, where the costs directly discourage a number of players. As a globally distributed GPU cloud, Aethir is perfectly positioned to solve this problem, providing high-performance, low-cost, on-demand decentralized Computing Power, which can not only steadily supply power to large companies but also help small and medium-sized AI teams to easily get started, spending less on unnecessary costs and focusing more on practical tasks. 2️⃣ AI is going global and cannot be limited by Computing Power: Aethir is not only the most important Computing Power provider, but also a bridge for Chinese AI to go global. Its decentralized architecture supports global deployment, allowing Chinese AI companies' products and services to truly reach overseas, rather than being constrained by local Computing Power. Of course, in the future, we also need to consider practical issues such as compliance and access models, but the trend is already very clear. 🎯 Summary: Aethir allows AI Computing Power to no longer be a bottleneck, empowering China's AI to showcase its strength on the global stage and letting good projects no longer be limited by GPU imagination. #BTC#

#Aethir: The AI-powered Computing Power engine that connects China with global innovation. 🔔 Introduction: Today's AI is no longer just a simple "question-answering" tool, but a multi-agent system capable of formulating plans and automatically executing tasks across platforms. China's AI is no longer a follower but is beginning to set its own standards. Local projects like Manus AI, Qwen, and ERNIE are demonstrating their strength and have already started sprinting in their respective tracks. 1️⃣ Computing Power Cost, the Biggest Barrier to AI Entrepreneurship: But no matter how strong AI is, without Computing Power it is useless. GPU prices are ridiculously high, becoming a "roadblock" for AI startups, especially in training large models, where the costs directly discourage a number of players. 2️⃣ AI is going global and cannot be limited by Computing Power: Aethir is not only the most important Computing Power provider, but also a bridge for Chinese AI to go global. Its decentralized architecture supports global deployment, allowing Chinese AI companies' products and services to truly reach overseas, rather than being constrained by local Computing Power. Of course, in the future, we also need to consider practical issues such as compliance and access models, but the trend is already very clear. 🎯 Summary: Aethir allows AI Computing Power to no longer be a bottleneck, helping China’s AI to showcase its strength on the global stage, letting great projects no longer be limited by GPU imagination. #AI ##Artificial Intelligence #创新 ##Intelligent

Ji Yichao, the founder of Manus, revealed on social platforms that Manus products use different fine-tuning models based on Ali's Qwen model. Manus, the first AIAgent product released by startup Monica, has recently gone viral on social media. (Interface)

$BABA #BABA Alibaba rolls out New Quark AI app in latest AI push Alibaba Group (BABA) has released the latest version of its Quark AI assistant mobile application, called New Quark, as part of its ongoing push to stay competitive in the rapidly growing AI market. New Quark uses Alibaba’s latest flagship Qwen AI model and integrates deep thinking, task execution and a chatbot. It also retains existing features such as image generation and travel planning. The new version will roll out in phases, with a pilot beginning on March 13. Competing apps include ByteDance’s (BDNCE) Doubao, Baidu’s (BIDU) (OTC:BAIDF) Ernie Bot, OpenAI's ChatGPT and Tencent (OTCPK:TCEHY) (OTCPK:TCTZF) Hunyuan. “This upgraded version of Quark is just the beginning. As our model capabilities continue to evolve, we envision Quark as a gateway to endless possibilities where users can explore everything with AI,” Wu Jia, CEO of Quark and Vice President of Alibaba Group said. A day earlier, Alibaba launched a new AI model, R1-Omni, which it said is capable of recognizing human emotions through audio and visual cues. Qwen has been benchmarked against DeepSeek, which triggered a race among Chinese and global AI companies. Alibaba itself has been pushing into the space rapidly with a major partnership with Apple for AI on iPhones and offering the R1-Omni for free on Hugging Face. The Chinese tech conglomerate has plans to increase investments in AI infrastructure, foundational models, AI-native applications and AI integration across businesses over the next three years.

Shenchao TechFlow News, on March 10, according to Jinshi Data, Manus founder Ji Yichao revealed on social platforms that Manus products use different fine-tuning models based on Ali's Qianwen large model (Qwen). Manus, the first AIAgent product released by startup Monica, has recently gone viral on social media.

99% of AI memes, worthless The remaining 99% of the AI memes are in the face of the rapid iteration of model capabilities such as Deepseek and Qwen, and the rapid development of agent systems such as Manus Living space is very limited Copying the bottom of ai meme is basically one in a million

The Alibaba Qwen team has officially released their latest research achievement - QwQ-32B large language model. With only about 1/21 of the parameter size of DeepSeek-R1, QwQ-32B has achieved a performance leap through reinforcement learning.