7*24

Hot topic

Market

Activities

According to monitoring by Dongcha Beating, the Claude Developer Console has launched the Prompt Caching Dashboard, accessible at platform.claude.com/usage/cache. The dashboard allows filtering by workspace, model, and time period, primarily displaying three key metrics: cache read ratio, which indicates the proportion of requests that hit existing cache; cache usage composition, which breaks down input tokens into four categories—uncached, 5-minute cache writes, 1-hour cache writes, and cache reads—presented as a stacked bar chart; and write amortization, which measures how many times a single cache write is reused by subsequent reads. In a screenshot example, Claude Opus 4.6 processed 2.74 billion input tokens over 7 days, achieving a read ratio of 85.4% and a write amortization of 8.65 times. Additionally, there is a time series graph of cache read ratios with granularity from 1 hour to 24 hours. Anthropic's prompt caching mechanism allows API users to mark system prompts, long contexts, and other fixed content as cacheable, charging an extra fee for the first write, while subsequent hits are billed at approximately one-tenth of the standard input price. The cache is retained by default for 5 minutes, with an option to extend it to 1 hour for a fee. Previously, users could only indirectly assess caching effectiveness through token count fields returned by the API, lacking a visual tool.

[Anthropic Launches Prompt Caching Dashboard for Visualizing Cache Hits and Costs]According to monitoring by Dongcha Beating, the Claude Developer Console has launched the Prompt Caching Dashboard, accessible at platform.claude.com/usage/cache. The dashboard allows filtering by workspace, model, and time period, primarily displaying three key metrics: cache read ratio, which indicates the proportion of requests that hit existing cache; cache usage composition, which breaks down input tokens into four categories—uncached, 5-minute cache writes, 1-hour cache writes, and cache reads—presented as a stacked bar chart; and write amortization, which measures how many times a single cache write is reused by subsequent reads. In a screenshot example, Claude Opus 4.6 processed 2.74 billion input tokens over 7 days, achieving a read ratio of 85.4% and a write amortization of 8.65 times. Additionally, there is a time series graph of cache read ratios with granularity from 1 hour to 24 hours. Anthropic's prompt caching mechanism allows API users to mark system prompts, long contexts, and other fixed content as cacheable, charging an extra fee for the first write, while subsequent hits are billed at approximately one-tenth of the standard input price. The cache is retained by default for 5 minutes, with an option to extend it to 1 hour for a fee. Previously, users could only indirectly assess caching effectiveness through token count fields returned by the API, lacking a visual tool.

Anthropic Launches Prompt Caching Dashboard for Visualizing Cache Hits and Costs

On April 22, shipping executives from commodity trading firms expressed concerns that Iran's toll collection in the Strait of Hormuz sets a dangerous precedent for the free flow of global trade. Since the outbreak of the Middle Eastern conflict at the end of February, Iran has attempted to impose transit fees on vessels passing through the strait, prompting strong protests from international shipping regulators. 'A highly dangerous precedent is being established here that undermines the passage rights of innocent vessels,' said Larry Johnson, global freight director at Mercuria Energy Trading. 'If some key passages or waterways begin to charge tolls, or are disrupted or threatened, creating a precedent for such fees, what will happen next? For instance, in places like the Black Sea, the Danish Strait, or the Strait of Malacca, who can predict? The free flow of global trade is no longer a reality.' (Jinshi)

[Opinion: Iran's Toll Collection Could Set a Dangerous Precedent]On April 22, shipping executives from commodity trading firms expressed concerns that Iran's toll collection in the Strait of Hormuz sets a dangerous precedent for the free flow of global trade. Since the outbreak of the Middle Eastern conflict at the end of February, Iran has attempted to impose transit fees on vessels passing through the strait, prompting strong protests from international shipping regulators. 'A highly dangerous precedent is being established here that undermines the passage rights of innocent vessels,' said Larry Johnson, global freight director at Mercuria Energy Trading. 'If some key passages or waterways begin to charge tolls, or are disrupted or threatened, creating a precedent for such fees, what will happen next? For instance, in places like the Black Sea, the Danish Strait, or the Strait of Malacca, who can predict? The free flow of global trade is no longer a reality.' (Jinshi)

Opinion: Iran's Toll Collection Could Set a Dangerous Precedent

On April 22, according to Artemis data, the circulation of USDT has surpassed $187 billion, setting a new historical record. Additionally, data from DefiLlama indicates that USDT's current market share stands at 58.82%.

[USDT Circulation Exceeds $187 Billion, Reaches All-Time High]On April 22, according to Artemis data, the circulation of USDT has surpassed $187 billion, setting a new historical record. Additionally, data from DefiLlama indicates that USDT's current market share stands at 58.82%.

USDT Circulation Exceeds $187 Billion, Reaches All-Time High

On April 22, Slow Fog reported that a highly destructive malware named 'MacSync Stealer' (v1.1.2) is currently active. This malware targets macOS users and steals sensitive data, including cryptocurrency wallets, browser credentials, system keychains, and infrastructure keys (SSH/AWS/K8s). The malware employs a spoofed AppleScript system dialog for phishing and displays a fake error message stating 'unsupported' after data leakage. It has immediately synchronized this IOC (Indicator of Compromise) with clients. Slow Fog advises users not to execute unverified macOS scripts and to remain highly vigilant against unexpected system password prompts. If an attack is suspected, immediate remediation is necessary: change all infrastructure credentials (SSH/AWS/K8s), invalidate any exposed keychains, and quickly migrate cryptocurrency assets to secure wallets.

[Slow Fog: Highly Destructive Malware on macOS Can Steal Sensitive Data Including Cryptocurrency Wallets]On April 22, Slow Fog reported that a highly destructive malware named 'MacSync Stealer' (v1.1.2) is currently active. This malware targets macOS users and steals sensitive data, including cryptocurrency wallets, browser credentials, system keychains, and infrastructure keys (SSH/AWS/K8s). The malware employs a spoofed AppleScript system dialog for phishing and displays a fake error message stating 'unsupported' after data leakage. It has immediately synchronized this IOC (Indicator of Compromise) with clients. Slow Fog advises users not to execute unverified macOS scripts and to remain highly vigilant against unexpected system password prompts. If an attack is suspected, immediate remediation is necessary: change all infrastructure credentials (SSH/AWS/K8s), invalidate any exposed keychains, and quickly migrate cryptocurrency assets to secure wallets.

Slow Fog: Highly Destructive Malware on macOS Can Steal Sensitive Data Including Cryptocurrency Wallets

According to monitoring by Dongcha Beating, Tim Frank, Vice President of AI Monetization at Microsoft, announced a set of updates to the commercial infrastructure aimed at the 'agentic web', designed to ensure that publishers, merchants, and advertisers continue to be discovered, cited, and transacted in scenarios where AI agents make decisions on behalf of users. The core consists of three parts. First is the Publisher Content Marketplace (PCM), which was first reported by Axios last year and is now being scaled up. PCM helps publishers receive compensation when their paid content is cited on AI platforms, covering non-free information such as map data, product catalogs, news, and health information. Copilot is the first demand-side participant, and Microsoft is negotiating access with other AI platforms. Unlike the one-time packaged payment RAG licensing agreements currently between publishers and AI platforms, PCM follows a continuous trading market approach. The second part is commercial agreement support. Microsoft has officially launched the Universal Commerce Protocol (UCP) data source in Merchant Center, an open standard developed by Google in collaboration with companies like Shopify, allowing AI agents to directly read structured product information. It also integrates Shopify's Global Catalog into Copilot, enabling over 500,000 merchants' products to be searched and transacted within Copilot. Microsoft claims that top Shopify merchants have seen a nearly 90% increase in exposure within Copilot after integration. Copilot Checkout has been extended to mobile and integrated into the membership systems of retailers like Target. The third part involves advertising and insights tools. The AI Visibility feature of Microsoft Clarity has been expanded, helping brands see how often their web pages are cited in AI responses, compare citations with competitors, and receive content gap suggestions. New offerings include AI Max search ads (in public beta since May), Offer Highlights (showcasing product selling points in Copilot conversations), and Audience generation (describing target audiences in natural language for AI to automatically generate audience targeting). Microsoft stated that it will not take transaction commissions, only charging a low technical service fee. Frank emphasized that the company has primarily focused on enterprise services for 51 years and does not have conflicts of interest as a platform competing with its customers.

[Microsoft Builds Commercial Infrastructure for the Era of AI Agents]According to monitoring by Dongcha Beating, Tim Frank, Vice President of AI Monetization at Microsoft, announced a set of updates to the commercial infrastructure aimed at the 'agentic web', designed to ensure that publishers, merchants, and advertisers continue to be discovered, cited, and transacted in scenarios where AI agents make decisions on behalf of users. The core consists of three parts. First is the Publisher Content Marketplace (PCM), which was first reported by Axios last year and is now being scaled up. PCM helps publishers receive compensation when their paid content is cited on AI platforms, covering non-free information such as map data, product catalogs, news, and health information. Copilot is the first demand-side participant, and Microsoft is negotiating access with other AI platforms. Unlike the one-time packaged payment RAG licensing agreements currently between publishers and AI platforms, PCM follows a continuous trading market approach. The second part is commercial agreement support. Microsoft has officially launched the Universal Commerce Protocol (UCP) data source in Merchant Center, an open standard developed by Google in collaboration with companies like Shopify, allowing AI agents to directly read structured product information. It also integrates Shopify's Global Catalog into Copilot, enabling over 500,000 merchants' products to be searched and transacted within Copilot. Microsoft claims that top Shopify merchants have seen a nearly 90% increase in exposure within Copilot after integration. Copilot Checkout has been extended to mobile and integrated into the membership systems of retailers like Target. The third part involves advertising and insights tools. The AI Visibility feature of Microsoft Clarity has been expanded, helping brands see how often their web pages are cited in AI responses, compare citations with competitors, and receive content gap suggestions. New offerings include AI Max search ads (in public beta since May), Offer Highlights (showcasing product selling points in Copilot conversations), and Audience generation (describing target audiences in natural language for AI to automatically generate audience targeting). Microsoft stated that it will not take transaction commissions, only charging a low technical service fee. Frank emphasized that the company has primarily focused on enterprise services for 51 years and does not have conflicts of interest as a platform competing with its customers.

Microsoft Builds Commercial Infrastructure for the Era of AI Agents

On April 22, QCP released a market outlook stating that BTC rebounded from an overnight low of around $75,000 to near $78,000. However, this trend reflects more of a relief in positions rather than a re-establishment of confidence. The market has reduced the risk of recent escalations following the extension of the ceasefire agreement in Iran, while Waller's testimony reinforced the Federal Reserve's reliance on data-driven decisions, without signaling a dovish shift. Risk sentiment has stabilized, but macro-level pressures remain firmly in place. For cryptocurrencies, BTC's rebound is driven more by a reduction in tail risks rather than fundamental improvements. Open interest has been rebuilt, and funding rates remain negative, indicating new short positions entering rather than long positions capitulating. This keeps the squeeze dynamics in play, but market conviction remains shallow. The options market has also signaled similarly. Front-end volatility is around 40 vol, which is still relatively low compared to realized volatility; skew remains tilted towards downside protection; and the term structure shows only a mild upward slope. Positioning continues to indicate a range-bound state rather than a sustained breakout. QCP believes that future paths remain anchored to oil prices and policy directions. A decline in oil prices or clearer signals from the Federal Reserve would support risk assets. In the absence of these factors, the market is likely to remain in a wait-and-see mode, pricing in uncertainty rather than seeking definitive solutions.

[QCP: Bitcoin's Recent Rebound Driven More by Risk Reduction than Fundamental Improvement]On April 22, QCP released a market outlook stating that BTC rebounded from an overnight low of around $75,000 to near $78,000. However, this trend reflects more of a relief in positions rather than a re-establishment of confidence. The market has reduced the risk of recent escalations following the extension of the ceasefire agreement in Iran, while Waller's testimony reinforced the Federal Reserve's reliance on data-driven decisions, without signaling a dovish shift. Risk sentiment has stabilized, but macro-level pressures remain firmly in place. For cryptocurrencies, BTC's rebound is driven more by a reduction in tail risks rather than fundamental improvements. Open interest has been rebuilt, and funding rates remain negative, indicating new short positions entering rather than long positions capitulating. This keeps the squeeze dynamics in play, but market conviction remains shallow. The options market has also signaled similarly. Front-end volatility is around 40 vol, which is still relatively low compared to realized volatility; skew remains tilted towards downside protection; and the term structure shows only a mild upward slope. Positioning continues to indicate a range-bound state rather than a sustained breakout. QCP believes that future paths remain anchored to oil prices and policy directions. A decline in oil prices or clearer signals from the Federal Reserve would support risk assets. In the absence of these factors, the market is likely to remain in a wait-and-see mode, pricing in uncertainty rather than seeking definitive solutions.

QCP: Bitcoin's Recent Rebound Driven More by Risk Reduction than Fundamental Improvement

According to monitoring by Beating, AI research lab NeoCognition has announced the completion of a $40 million seed round financing, emerging from stealth mode. The company is headquartered in Palo Alto, California, and was co-founded by Ohio State University associate professor Yu Su along with Xiang Deng and Yu Gu. This round was oversubscribed, with Cambium Capital and Walden Catalyst Ventures leading the investment, and Vista Equity Partners participating. Angel investors include Intel CEO Lip-Bu Tan, Databricks co-founder and executive chairman Ion Stoica, as well as AI researchers Dawn Song, Ruslan Salakhutdinov, and Luke Zettlemoyer. NeoCognition aims to address the reliability of AI agents. Founder Yu Su stated that the current success rate of AI agents completing tasks as expected is only about 50%, making each execution feel like a roll of the dice. The company's approach is to enable agents to autonomously build a 'world model' in their respective fields through actual use, capturing the rules, relationships, and constraints of specific business environments, allowing for rapid specialization on the job rather than relying on generic pre-training. Before ChatGPT, Yu Su established an LLM agent research lab at Ohio State University, with the team's previous notable works including Mind2Web, MMMU, and SeeAct. The business model targets the enterprise sector, primarily serving SaaS companies by helping them embed continuously improving AI employees into their own products and operational processes. Vista Equity Partners, which holds a significant number of enterprise software companies, views its investment as a distribution channel. The team consists of about 15 members, most of whom hold doctoral degrees, and there are currently no publicly available products.

[NeoCognition Secures $40 Million Seed Round for Specialized AI Agents Capable of On-the-Job Learning]According to monitoring by Beating, AI research lab NeoCognition has announced the completion of a $40 million seed round financing, emerging from stealth mode. The company is headquartered in Palo Alto, California, and was co-founded by Ohio State University associate professor Yu Su along with Xiang Deng and Yu Gu. This round was oversubscribed, with Cambium Capital and Walden Catalyst Ventures leading the investment, and Vista Equity Partners participating. Angel investors include Intel CEO Lip-Bu Tan, Databricks co-founder and executive chairman Ion Stoica, as well as AI researchers Dawn Song, Ruslan Salakhutdinov, and Luke Zettlemoyer. NeoCognition aims to address the reliability of AI agents. Founder Yu Su stated that the current success rate of AI agents completing tasks as expected is only about 50%, making each execution feel like a roll of the dice. The company's approach is to enable agents to autonomously build a 'world model' in their respective fields through actual use, capturing the rules, relationships, and constraints of specific business environments, allowing for rapid specialization on the job rather than relying on generic pre-training. Before ChatGPT, Yu Su established an LLM agent research lab at Ohio State University, with the team's previous notable works including Mind2Web, MMMU, and SeeAct. The business model targets the enterprise sector, primarily serving SaaS companies by helping them embed continuously improving AI employees into their own products and operational processes. Vista Equity Partners, which holds a significant number of enterprise software companies, views its investment as a distribution channel. The team consists of about 15 members, most of whom hold doctoral degrees, and there are currently no publicly available products.

NeoCognition Secures $40 Million Seed Round for Specialized AI Agents Capable of On-the-Job Learning

On April 22, Slow Fog reported that a highly destructive malware named "MacSync Stealer" (v1.1.2) is currently active. This malware targets macOS users, stealing sensitive data including cryptocurrency wallets, browser credentials, system keychains, and infrastructure keys (SSH/AWS/K8s). It employs a spoofed AppleScript system dialog for phishing and displays a fake error message stating "unsupported" after data leakage. The IOC (Indicator of Compromise) has been promptly synchronized with clients. Slow Fog advises users not to execute unverified macOS scripts and to remain highly vigilant against unexpected system password prompts. If an attack is suspected, immediate remediation is necessary: change all infrastructure credentials (SSH/AWS/K8s), invalidate exposed keychains, and quickly migrate cryptocurrency assets to secure wallets.

[Slow Fog: Highly Destructive Malware on macOS Can Steal Sensitive Data Including Cryptocurrency Wallets]On April 22, Slow Fog reported that a highly destructive malware named "MacSync Stealer" (v1.1.2) is currently active. This malware targets macOS users, stealing sensitive data including cryptocurrency wallets, browser credentials, system keychains, and infrastructure keys (SSH/AWS/K8s). It employs a spoofed AppleScript system dialog for phishing and displays a fake error message stating "unsupported" after data leakage. The IOC (Indicator of Compromise) has been promptly synchronized with clients. Slow Fog advises users not to execute unverified macOS scripts and to remain highly vigilant against unexpected system password prompts. If an attack is suspected, immediate remediation is necessary: change all infrastructure credentials (SSH/AWS/K8s), invalidate exposed keychains, and quickly migrate cryptocurrency assets to secure wallets.

On April 22, UK investors can once again gain the opportunity to purchase cryptocurrency exchange-traded notes (ETNs) tax-free through the Innovative Finance Individual Savings Account (ISA) channel.

[UK Investors Can Access ETNs Through ISA Channel]On April 22, UK investors can once again gain the opportunity to purchase cryptocurrency exchange-traded notes (ETNs) tax-free through the Innovative Finance Individual Savings Account (ISA) channel.

UK Investors Can Access ETNs Through ISA Channel

Blockchain networking and security company Syntropy has raised $4m in a post-seed funding round led by Alpha Transform Group, Alphemy Capital, Denali Digital, Saxon Ventures, and FirstBlock Capital.

[Syntropy Raises $4 Million to Unlock the Web3 Potential]Blockchain networking and security company Syntropy has raised $4m in a post-seed funding round led by Alpha Transform Group, Alphemy Capital, Denali Digital, Saxon Ventures, and FirstBlock Capital.

Syntropy Raises $4 Million to Unlock the Web3 Potential

Bitcoin Created By The CIA? Chinese Professor Jiang Xueqin Makes Bold Claim

BIT Research: Five Signals Flash Simultaneously, Has the Bitcoin Bear Market Ended?

Bitcoin Recovery Fails To Lift Market Sentiment From Extreme Fear

Nic Carter Says Bitcoin Has 3 Ways To Handle Satoshi’s Coins

Crypto And Financial Industry Giants Reveal What X Money Launch Means

Crypto Regulation: Polish Parliament Fails To Overturn Presidential Veto Again

Bitcoin Price Could See Another Crash, But What Is The Long-Term Prognosis?

BTC Market Pulse: Week 17

Shiba Inu Crosses 20,000 Burn Transactions Milestone, Dogecoin Eyes X Money, But Why Are Prices Down?

Strategy Surpasses 800K BTC as $2.5B Bitcoin Buying Spree Continues

Article

TRX fell below 0.33  USDT

Syntropy Raises $4 Million to Unlock the Web3 Potential

Descargo de responsabilidad：El contenido anterior no representa las posiciones de HTX.，HTX no ofrece ninguna recomendación de trading.。

Todos los comentarios0Lo más recientePopular

Temas populares