7*24

Hot topic

Market

Activities

On April 15, former U.S. Treasury Secretary Janet Yellen stated that Trump's pressure on the Federal Reserve to lower interest rates in order to reduce U.S. debt costs is akin to 'banana republic' rhetoric. Speaking at HSBC's Global Investment Summit, Yellen issued a warning regarding the independence of monetary policy, saying she had 'never seen such a level of threat to the Federal Reserve.' She questioned, 'Does a president of a developed country often express the view that interest rates should be set to lower debt servicing costs?' She added, 'This is common in banana republics.' She noted that managing interest rates for the sake of government budgets has led to 'hyperinflation' in some countries.

[Former U.S. Treasury Secretary Yellen: Trump's Push to Lower U.S. Interest Rates is 'Banana Republic' Behavior]On April 15, former U.S. Treasury Secretary Janet Yellen stated that Trump's pressure on the Federal Reserve to lower interest rates in order to reduce U.S. debt costs is akin to 'banana republic' rhetoric. Speaking at HSBC's Global Investment Summit, Yellen issued a warning regarding the independence of monetary policy, saying she had 'never seen such a level of threat to the Federal Reserve.' She questioned, 'Does a president of a developed country often express the view that interest rates should be set to lower debt servicing costs?' She added, 'This is common in banana republics.' She noted that managing interest rates for the sake of government budgets has led to 'hyperinflation' in some countries.

Former U.S. Treasury Secretary Yellen: Trump's Push to Lower U.S. Interest Rates is 'Banana Republic' Behavior

On April 15, former U.S. Treasury Secretary Janet Yellen stated that Federal Reserve Chair nominee Kevin Walsh lacks 'credibility' in advocating for interest rate cuts. (Jin Shi)

[Former Treasury Secretary Yellen: Fed Chair Nominee Kevin Walsh Lacks Credibility on Rate Cuts]On April 15, former U.S. Treasury Secretary Janet Yellen stated that Federal Reserve Chair nominee Kevin Walsh lacks 'credibility' in advocating for interest rate cuts. (Jin Shi)

Former Treasury Secretary Yellen: Fed Chair Nominee Kevin Walsh Lacks Credibility on Rate Cuts

According to monitoring by Dongcha Beating, Hugging Face CEO Clem Delangue announced the official launch of Kernels on Hub. GPU operators are low-level optimization codes that allow graphics cards to achieve extreme speeds, accelerating inference and training by 1.7 to 2.5 times. However, installation has been a nightmare: for example, the commonly used FlashAttention requires about 96GB of memory and several hours for local compilation, and even slight mismatches in PyTorch or CUDA versions can lead to errors, causing most developers to get stuck at this installation step. Kernels Hub moves the compilation to the cloud. Hugging Face pre-compiles the operators for various graphics cards and system environments, allowing developers to write one line of code, with the Hub automatically matching the hardware environment and downloading the pre-compiled files for immediate use within seconds. The same process can load multiple different versions of operators and is compatible with torch.compile. Kernels was tested and launched in June last year and has now been upgraded to a first-level repository type on Hub, alongside Models, Datasets, and Spaces. Currently, there are 61 pre-compiled operators covering common scenarios such as attention mechanisms, normalization, mixture of experts routing, and quantization, supporting four hardware acceleration platforms: NVIDIA CUDA, AMD ROCm, Apple Metal, and Intel XPU. It has been integrated into Hugging Face's inference framework TGI and Transformers library.

[Hugging Face Launches Kernels, GPU Operators Installed with One Line of Code]According to monitoring by Dongcha Beating, Hugging Face CEO Clem Delangue announced the official launch of Kernels on Hub. GPU operators are low-level optimization codes that allow graphics cards to achieve extreme speeds, accelerating inference and training by 1.7 to 2.5 times. However, installation has been a nightmare: for example, the commonly used FlashAttention requires about 96GB of memory and several hours for local compilation, and even slight mismatches in PyTorch or CUDA versions can lead to errors, causing most developers to get stuck at this installation step. Kernels Hub moves the compilation to the cloud. Hugging Face pre-compiles the operators for various graphics cards and system environments, allowing developers to write one line of code, with the Hub automatically matching the hardware environment and downloading the pre-compiled files for immediate use within seconds. The same process can load multiple different versions of operators and is compatible with torch.compile. Kernels was tested and launched in June last year and has now been upgraded to a first-level repository type on Hub, alongside Models, Datasets, and Spaces. Currently, there are 61 pre-compiled operators covering common scenarios such as attention mechanisms, normalization, mixture of experts routing, and quantization, supporting four hardware acceleration platforms: NVIDIA CUDA, AMD ROCm, Apple Metal, and Intel XPU. It has been integrated into Hugging Face's inference framework TGI and Transformers library.

Hugging Face Launches Kernels, GPU Operators Installed with One Line of Code

On April 15, according to Coinglass data, RAVE experienced liquidations totaling $30.6 million in the past 24 hours, second only to BTC and ETH. Among these, long positions accounted for $6.93 million in liquidations, while short positions saw $23.67 million. Currently, the average funding rate for RAVE across major platforms is as high as -1.3695%. As the price of the coin continues to rise, more and more investors are opting to enter the market to short.

[RAVE Liquidation Volume Second Only to BTC and ETH, Investor Short Sentiment Soars]On April 15, according to Coinglass data, RAVE experienced liquidations totaling $30.6 million in the past 24 hours, second only to BTC and ETH. Among these, long positions accounted for $6.93 million in liquidations, while short positions saw $23.67 million. Currently, the average funding rate for RAVE across major platforms is as high as -1.3695%. As the price of the coin continues to rise, more and more investors are opting to enter the market to short.

RAVE Liquidation Volume Second Only to BTC and ETH, Investor Short Sentiment Soars

According to monitoring by Dongcha Beating, Zhiyu raised the prices of the entire GLM Coding Plan overseas version by 80%-140% on April 11. The Max monthly fee increased from $80 to $160, Pro from $30 to $72, and Lite from $10 to $18 (all prices include a 10% invitation discount). The same Max package is priced at 469 yuan (approximately $68) domestically, making the overseas version more than double the domestic price after the increase. Additionally, with the release of GLM-5.1, the API prices were adjusted by 10%, marking the third price increase by Zhiyu in 2026. Overseas developer Lex posted on X, stating that this 'foreigner tax' is outrageous. He found that by paying through WeChat or Alipay, one can purchase the domestic version, and the API key obtained can also be used overseas without any regional lock, indicating that the price difference is purely a pricing strategy. The comments section saw polarized debates: opponents argued that all overseas endpoints are actually hosted in Singapore, and the infrastructure costs are not higher; supporters claimed that the payment, risk control, and compliance costs in overseas markets should be borne by local users, and regional differential pricing is an industry norm. The continuous price increases are backed by product reputation. Victor Mustar, product head at Hugging Face, posted on X that he wrote a 531-line Three.js racing game from scratch using GLM-5.1, which includes a real drifting physics engine and four AI personalities, with no visual aids and the model self-iterating and debugging over 20 times, believing its performance surpasses that of Claude Code combined with Opus 4.6, and predicting that this model will be widely discussed. However, the price increase comes at a cost. OpenRouter data shows that the GLM series ranking has gradually declined from third globally when GLM-5 was released in February, dropping to 15th place by April 15. The issue with the domestic version is not its price, but its availability; the Max package sells out instantly each day, and reselling services have appeared on Xianyu.

[Zhiyu GLM Overseas Version Prices Increase by Up to 140%, Price Difference with Domestic Version Exceeds Double]According to monitoring by Dongcha Beating, Zhiyu raised the prices of the entire GLM Coding Plan overseas version by 80%-140% on April 11. The Max monthly fee increased from $80 to $160, Pro from $30 to $72, and Lite from $10 to $18 (all prices include a 10% invitation discount). The same Max package is priced at 469 yuan (approximately $68) domestically, making the overseas version more than double the domestic price after the increase. Additionally, with the release of GLM-5.1, the API prices were adjusted by 10%, marking the third price increase by Zhiyu in 2026. Overseas developer Lex posted on X, stating that this 'foreigner tax' is outrageous. He found that by paying through WeChat or Alipay, one can purchase the domestic version, and the API key obtained can also be used overseas without any regional lock, indicating that the price difference is purely a pricing strategy. The comments section saw polarized debates: opponents argued that all overseas endpoints are actually hosted in Singapore, and the infrastructure costs are not higher; supporters claimed that the payment, risk control, and compliance costs in overseas markets should be borne by local users, and regional differential pricing is an industry norm. The continuous price increases are backed by product reputation. Victor Mustar, product head at Hugging Face, posted on X that he wrote a 531-line Three.js racing game from scratch using GLM-5.1, which includes a real drifting physics engine and four AI personalities, with no visual aids and the model self-iterating and debugging over 20 times, believing its performance surpasses that of Claude Code combined with Opus 4.6, and predicting that this model will be widely discussed. However, the price increase comes at a cost. OpenRouter data shows that the GLM series ranking has gradually declined from third globally when GLM-5 was released in February, dropping to 15th place by April 15. The issue with the domestic version is not its price, but its availability; the Max package sells out instantly each day, and reselling services have appeared on Xianyu.

Zhiyu GLM Overseas Version Prices Increase by Up to 140%, Price Difference with Domestic Version Exceeds Double

On April 15, Jamie Dimon, the CEO of JPMorgan Chase (JPM.N), stated on Tuesday that while artificial intelligence tools may ultimately help businesses defend against cyberattacks, at the current stage, they actually make companies more vulnerable. "AI makes things worse and more difficult," Dimon told analysts during a conference call on Tuesday morning. "It does bring additional vulnerabilities, and perhaps in the long run, it will provide better ways to fortify oneself." These remarks indicate that AI, which companies view as a technology to enhance productivity, has also evolved into a serious threat, as it provides criminals with new methods to infiltrate technical systems. Last week, U.S. Treasury Secretary Janet Yellen and Federal Reserve Chairman Jerome Powell convened bank CEOs to discuss the risks posed by Anthropic's cutting-edge AI model, Mythos, designed specifically for defensive cybersecurity workflows.

[JPMorgan CEO: AI is Not Just a Tool, But Also a Threat]On April 15, Jamie Dimon, the CEO of JPMorgan Chase (JPM.N), stated on Tuesday that while artificial intelligence tools may ultimately help businesses defend against cyberattacks, at the current stage, they actually make companies more vulnerable. "AI makes things worse and more difficult," Dimon told analysts during a conference call on Tuesday morning. "It does bring additional vulnerabilities, and perhaps in the long run, it will provide better ways to fortify oneself." These remarks indicate that AI, which companies view as a technology to enhance productivity, has also evolved into a serious threat, as it provides criminals with new methods to infiltrate technical systems. Last week, U.S. Treasury Secretary Janet Yellen and Federal Reserve Chairman Jerome Powell convened bank CEOs to discuss the risks posed by Anthropic's cutting-edge AI model, Mythos, designed specifically for defensive cybersecurity workflows.

JPMorgan CEO: AI is Not Just a Tool, But Also a Threat

On April 15, former U.S. Treasury Secretary Janet Yellen stated that despite the inflationary pressures on the global economy caused by supply shocks from the Iran conflict, the Federal Reserve may still implement a rate cut this year. Speaking at the HSBC Global Investment Summit in Hong Kong on Wednesday, Yellen noted, "Short-term inflation expectations have risen slightly, but the Fed will be very cautious in observing all data, and I believe they are keeping an open mind." If a judgment were to be made at the next Federal Open Market Committee (FOMC) meeting that produces a forecast report, Yellen believes there could be a rate cut later this year. The Fed policymakers decided in March to maintain the benchmark interest rate in the range of 3.5% to 3.75%. At that time, most officials expected that at least one rate cut would be appropriate this year. Former President Trump has repeatedly criticized Powell for not implementing the significant rate cuts he desired. Trump expressed confidence that his nominee for Powell's successor, Kevin Warsh, would carry out those rate cuts. Yellen emphasized that the conflict in the Middle East has heightened economic uncertainty, stating, "This has put upward pressure on inflation, which we have already seen in recent inflation reports, and we may see more in the future. This is essentially a broad supply shock."

[Former Treasury Secretary Yellen: Fed May Cut Rates Once This Year]On April 15, former U.S. Treasury Secretary Janet Yellen stated that despite the inflationary pressures on the global economy caused by supply shocks from the Iran conflict, the Federal Reserve may still implement a rate cut this year. Speaking at the HSBC Global Investment Summit in Hong Kong on Wednesday, Yellen noted, "Short-term inflation expectations have risen slightly, but the Fed will be very cautious in observing all data, and I believe they are keeping an open mind." If a judgment were to be made at the next Federal Open Market Committee (FOMC) meeting that produces a forecast report, Yellen believes there could be a rate cut later this year. The Fed policymakers decided in March to maintain the benchmark interest rate in the range of 3.5% to 3.75%. At that time, most officials expected that at least one rate cut would be appropriate this year. Former President Trump has repeatedly criticized Powell for not implementing the significant rate cuts he desired. Trump expressed confidence that his nominee for Powell's successor, Kevin Warsh, would carry out those rate cuts. Yellen emphasized that the conflict in the Middle East has heightened economic uncertainty, stating, "This has put upward pressure on inflation, which we have already seen in recent inflation reports, and we may see more in the future. This is essentially a broad supply shock."

Former Treasury Secretary Yellen: Fed May Cut Rates Once This Year

On April 15, data showed that Anthropic's valuation in the Pre-IPO market segment on Hyperliquid has surpassed that of OpenAI, currently priced at $934, corresponding to a valuation of $934 billion, with a 24-hour increase of 6%. In contrast, OpenAI is currently priced at $824, corresponding to a valuation of $824 billion, with a 24-hour decrease of 6.3%. However, the market currently has very low trading volume, with Anthropic's 24-hour trading volume at only $372,000 and OpenAI's at just $123,000.

[Anthropic Surpasses OpenAI in Valuation on Hyperliquid, Now at $934 Billion]On April 15, data showed that Anthropic's valuation in the Pre-IPO market segment on Hyperliquid has surpassed that of OpenAI, currently priced at $934, corresponding to a valuation of $934 billion, with a 24-hour increase of 6%. In contrast, OpenAI is currently priced at $824, corresponding to a valuation of $824 billion, with a 24-hour decrease of 6.3%. However, the market currently has very low trading volume, with Anthropic's 24-hour trading volume at only $372,000 and OpenAI's at just $123,000.

Anthropic Surpasses OpenAI in Valuation on Hyperliquid, Now at $934 Billion

On April 15, the U.S. Securities and Exchange Commission (SEC) announced the repeal of the 'Pattern Day Trader' rule, eliminating the requirement for day trading accounts to maintain a minimum balance of $25,000. This rule was originally intended to restrict frequent trading behavior and reduce investor risk. The adjustment is expected to impact the trading thresholds of various financial products, including cryptocurrencies and Web3-related assets, providing individual investors with greater flexibility. The SEC's move signifies a further relaxation of the U.S. financial regulatory environment, which may enhance market activity.

[SEC Abolishes 'Pattern Day Trader' Rule, Removes Minimum Balance Requirement]On April 15, the U.S. Securities and Exchange Commission (SEC) announced the repeal of the 'Pattern Day Trader' rule, eliminating the requirement for day trading accounts to maintain a minimum balance of $25,000. This rule was originally intended to restrict frequent trading behavior and reduce investor risk. The adjustment is expected to impact the trading thresholds of various financial products, including cryptocurrencies and Web3-related assets, providing individual investors with greater flexibility. The SEC's move signifies a further relaxation of the U.S. financial regulatory environment, which may enhance market activity.

SEC Abolishes 'Pattern Day Trader' Rule, Removes Minimum Balance Requirement

The recent Shapella upgrade on the Ethereum network has led to a surge in deposit activity, with ether holders rushing to stake their tokens with network validators. This has resulted in the total volume of crypto-asset staked exceeding 20 million, a significant milestone.

[Ether Holders Rush to Stake Tokens as Total Volume of Crypto-Asset Staked Surpasses 20 Million]The recent Shapella upgrade on the Ethereum network has led to a surge in deposit activity, with ether holders rushing to stake their tokens with network validators. This has resulted in the total volume of crypto-asset staked exceeding 20 million, a significant milestone.

Ether Holders Rush to Stake Tokens as Total Volume of Crypto-Asset Staked Surpasses 20 Million

Article

Tencent Cloud Supports One-Click Deployment of Hermes Agent

BTC broke through 76000  USDT

CoW Swap Frontend Issues, Users Advised to Refrain from Use

Cloud Computing Startup Fluidstack Seeks to Raise $1 Billion at $18 Billion Valuation

Aave: CowSwap Frontend Attack Incident Does Not Affect Aave Interface or Underlying Protocol Security

Central Bank of Russia Proposes Mandatory Real-Name System for Crypto Transactions to Enhance Fund Flow Supervision

Ether Holders Rush to Stake Tokens as Total Volume of Crypto-Asset Staked Surpasses 20 Million

Disclaimer：The content above does not represent HTX's positions.，HTX does not provide any trading recommendations.。

All Comments0LatestHot

Trending Topics