7*24

Hot topic

Market

Activities

On May 26, Strive ($SATA), a Nasdaq-listed Bitcoin treasury company, reached a 3-month high, soaring to $100.53.

[Strive Stock Price Surpasses $100, Reaches 3-Month High]On May 26, Strive ($SATA), a Nasdaq-listed Bitcoin treasury company, reached a 3-month high, soaring to $100.53.

Strive Stock Price Surpasses $100, Reaches 3-Month High

On May 26, the Islamic Republic of Iran Broadcasting reported that some foreign media claimed Iran and the United States had reached a 14-point understanding memorandum, which is "completely fabricated and baseless." The report stated that some media outlets had claimed hours earlier to have obtained the final draft of the Iran-U.S. understanding memorandum, which allegedly included issues such as opening the Strait of Hormuz, initiating nuclear negotiations, and unfreezing Iran's frozen assets. However, these claims are not true.

[Iranian State Media Denies Understanding Memorandum with the U.S.]On May 26, the Islamic Republic of Iran Broadcasting reported that some foreign media claimed Iran and the United States had reached a 14-point understanding memorandum, which is "completely fabricated and baseless." The report stated that some media outlets had claimed hours earlier to have obtained the final draft of the Iran-U.S. understanding memorandum, which allegedly included issues such as opening the Strait of Hormuz, initiating nuclear negotiations, and unfreezing Iran's frozen assets. However, these claims are not true.

Iranian State Media Denies Understanding Memorandum with the U.S.

According to Beating Monitoring, Mianbi Intelligence and Tsinghua NLP Laboratory have jointly open-sourced the world's first production-level pre-training framework, ForgeTrain, entirely written by AI, within the OpenBMB community. They also released MiniCPM5-1B, a small model trained using ForgeTrain. As the first example demonstrating a closed-loop 'AI manufacturing AI' project, ForgeTrain outperforms NVIDIA's Megatron under the same hardware conditions and achieves a 10% acceleration when pre-trained on Huawei's Ascend. Additionally, MiniCPM5-1B topped the Artificial Analysis leaderboard for open-weight small models. To enable AI to autonomously build foundational pre-training infrastructure, Mianbi Intelligence proposed the 'Forge Engineering' software programming paradigm, moving away from a universal framework compatible with all hardware and tasks, and instead leveraging AI's low-cost code generation capabilities to forge dedicated code for specific models and hardware on-site. In terms of construction mechanism, ForgeTrain employs a three-stage approach: first, it collects key data from existing pre-training frameworks to form a testing harness; next, it iteratively generates binary-consistent framework code in an automated closed loop; and finally, it lifts restrictions to surpass the reference implementation. This entire automated evolution corresponds to the L3 to L4 stages of AI manufacturing AI. As the first output model of ForgeTrain, MiniCPM5-1B features 1.08 billion parameters, with a core architecture based on the standard LlamaForCausalLM design, significantly lowering the barriers for downstream integration and inference deployment. In the Artificial Analysis evaluation, the model scored 18, surpassing the 2B scale Qwen3.5-2B (16 points), and leading over Qwen3.5-0.8B (11 points) and LFM2.5-1.2B-Thinking (8 points). The model supports deployment formats such as MLX 4-bit and GGUF Q4_K_M, with INT4 quantized weights of only 0.5GB, and natively supports long text contexts of 131,072 tokens along with mixed dual-mode inference based on enable_thinking. Leveraging extremely low hardware overhead, OpenBMB has also open-sourced a purely offline desktop companion application, MiniCPM Desk Pet, which supports real-time responses to coding activities in development tools like Cursor and LoRA character switching.

[World's First AI-Written Pre-Training Framework Open-Sourced: ForgeTrain by Tsinghua and Mianbi]According to Beating Monitoring, Mianbi Intelligence and Tsinghua NLP Laboratory have jointly open-sourced the world's first production-level pre-training framework, ForgeTrain, entirely written by AI, within the OpenBMB community. They also released MiniCPM5-1B, a small model trained using ForgeTrain. As the first example demonstrating a closed-loop 'AI manufacturing AI' project, ForgeTrain outperforms NVIDIA's Megatron under the same hardware conditions and achieves a 10% acceleration when pre-trained on Huawei's Ascend. Additionally, MiniCPM5-1B topped the Artificial Analysis leaderboard for open-weight small models. To enable AI to autonomously build foundational pre-training infrastructure, Mianbi Intelligence proposed the 'Forge Engineering' software programming paradigm, moving away from a universal framework compatible with all hardware and tasks, and instead leveraging AI's low-cost code generation capabilities to forge dedicated code for specific models and hardware on-site. In terms of construction mechanism, ForgeTrain employs a three-stage approach: first, it collects key data from existing pre-training frameworks to form a testing harness; next, it iteratively generates binary-consistent framework code in an automated closed loop; and finally, it lifts restrictions to surpass the reference implementation. This entire automated evolution corresponds to the L3 to L4 stages of AI manufacturing AI. As the first output model of ForgeTrain, MiniCPM5-1B features 1.08 billion parameters, with a core architecture based on the standard LlamaForCausalLM design, significantly lowering the barriers for downstream integration and inference deployment. In the Artificial Analysis evaluation, the model scored 18, surpassing the 2B scale Qwen3.5-2B (16 points), and leading over Qwen3.5-0.8B (11 points) and LFM2.5-1.2B-Thinking (8 points). The model supports deployment formats such as MLX 4-bit and GGUF Q4_K_M, with INT4 quantized weights of only 0.5GB, and natively supports long text contexts of 131,072 tokens along with mixed dual-mode inference based on enable_thinking. Leveraging extremely low hardware overhead, OpenBMB has also open-sourced a purely offline desktop companion application, MiniCPM Desk Pet, which supports real-time responses to coding activities in development tools like Cursor and LoRA character switching.

World's First AI-Written Pre-Training Framework Open-Sourced: ForgeTrain by Tsinghua and Mianbi

On May 26, Bitcoin's expected volatility significantly dropped to a nine-month low. The Bitcoin VIX index, which reflects the 30-day expected volatility based on options pricing, fell to 36.11, the lowest since September of last year, and is approaching the lowest point of 2023. Diminished trading activity, coupled with speculative funds shifting from Bitcoin to other assets, has led to a reduced demand for options protection. Bitcoin's price has also retreated to levels near those seen since September of last year, resulting in an overall 'breather' in the crypto market. Market sentiment remains relatively calm, and volatility is expected to stay low in the short term.

[Bitcoin Volatility Nears 2023 Low as Crypto Market Enters 'Calm Breather']On May 26, Bitcoin's expected volatility significantly dropped to a nine-month low. The Bitcoin VIX index, which reflects the 30-day expected volatility based on options pricing, fell to 36.11, the lowest since September of last year, and is approaching the lowest point of 2023. Diminished trading activity, coupled with speculative funds shifting from Bitcoin to other assets, has led to a reduced demand for options protection. Bitcoin's price has also retreated to levels near those seen since September of last year, resulting in an overall 'breather' in the crypto market. Market sentiment remains relatively calm, and volatility is expected to stay low in the short term.

Bitcoin Volatility Nears 2023 Low as Crypto Market Enters 'Calm Breather'

Glassnode tweeted that at the current price, approximately 7.75 million BTC are in a state of loss. This oversupply is a structural characteristic of a bear market and is typically resolved only when weak hands capitulate.

[Glassnode: Approximately 7.75 Million BTC Currently in Loss]Glassnode tweeted that at the current price, approximately 7.75 million BTC are in a state of loss. This oversupply is a structural characteristic of a bear market and is typically resolved only when weak hands capitulate.

Glassnode: Approximately 7.75 Million BTC Currently in Loss

On May 26, CoinShares released its latest weekly report, indicating that digital asset investment products experienced a net outflow of $1.47 billion last week, marking the second consecutive week of negative growth and the third-largest weekly outflow in 2026. The total outflow over the past two weeks has reached $2.54 billion. By asset type, Bitcoin saw an outflow of $1.315 billion, the largest weekly outflow in 2026, reducing its year-to-date inflow from $3.9 billion to $2.6 billion. Ethereum recorded an outflow of $223 million, remaining roughly stable compared to the previous week. Some altcoins still saw slight inflows, with XRP gaining $31.8 million, Near $9 million, and Solana $7.7 million. By region, the United States dominated the outflow pattern with a weekly outflow of $1.425 billion, while Switzerland, Canada, and Hong Kong saw outflows of $16.2 million, $12.5 million, and $12.2 million, respectively. Risk aversion sentiment has spread from localized areas to most regions globally. CoinShares noted that the outflows are closely related to rising geopolitical risks associated with Iran. Despite the ongoing legislative progress of the CLARITY Act, market risk aversion continues to deepen.

[Digital Asset Investment Products See $1.47 Billion Net Outflow Last Week, Bitcoin Records Largest Weekly Outflow of $1.315 Billion in 2026]On May 26, CoinShares released its latest weekly report, indicating that digital asset investment products experienced a net outflow of $1.47 billion last week, marking the second consecutive week of negative growth and the third-largest weekly outflow in 2026. The total outflow over the past two weeks has reached $2.54 billion. By asset type, Bitcoin saw an outflow of $1.315 billion, the largest weekly outflow in 2026, reducing its year-to-date inflow from $3.9 billion to $2.6 billion. Ethereum recorded an outflow of $223 million, remaining roughly stable compared to the previous week. Some altcoins still saw slight inflows, with XRP gaining $31.8 million, Near $9 million, and Solana $7.7 million. By region, the United States dominated the outflow pattern with a weekly outflow of $1.425 billion, while Switzerland, Canada, and Hong Kong saw outflows of $16.2 million, $12.5 million, and $12.2 million, respectively. Risk aversion sentiment has spread from localized areas to most regions globally. CoinShares noted that the outflows are closely related to rising geopolitical risks associated with Iran. Despite the ongoing legislative progress of the CLARITY Act, market risk aversion continues to deepen.

Digital Asset Investment Products See $1.47 Billion Net Outflow Last Week, Bitcoin Records Largest Weekly Outflow of $1.315 Billion in 2026

On May 26, Reuters reported that Isabel Schnabel, a member of the European Central Bank's governing council and a potential candidate for ECB president, stated that even if the U.S. and Iran achieve results in peace negotiations, the ECB should still initiate a rate hike in June. She believes that the current energy shock is lasting much longer than expected, and damage to energy infrastructure along with supply chain disruptions have pushed high energy prices into broader economic sectors. Schnabel pointed out that inflation in the Eurozone has risen to 3%, with further upward risks still present. There are concerns within the ECB that high energy costs are driving up prices for goods and services through 'second-round effects,' creating more persistent inflationary pressures. She emphasized that the ECB's previous baseline forecast included two rate hikes, and a single action may not be sufficient to achieve the inflation control target.

[ECB Governing Council Member: Rate Hike Still Needed in June Even if Iran Reaches Peace Agreement]On May 26, Reuters reported that Isabel Schnabel, a member of the European Central Bank's governing council and a potential candidate for ECB president, stated that even if the U.S. and Iran achieve results in peace negotiations, the ECB should still initiate a rate hike in June. She believes that the current energy shock is lasting much longer than expected, and damage to energy infrastructure along with supply chain disruptions have pushed high energy prices into broader economic sectors. Schnabel pointed out that inflation in the Eurozone has risen to 3%, with further upward risks still present. There are concerns within the ECB that high energy costs are driving up prices for goods and services through 'second-round effects,' creating more persistent inflationary pressures. She emphasized that the ECB's previous baseline forecast included two rate hikes, and a single action may not be sufficient to achieve the inflation control target.

ECB Governing Council Member: Rate Hike Still Needed in June Even if Iran Reaches Peace Agreement

According to monitoring by Dongcha Beating, Alipay has announced the launch of the world's first Token Pay service and AI wallet product. Today, the company's full-stack AI payment product matrix has been unveiled, marking the establishment of the world's first large-scale commercial AI-native payment infrastructure. Among these, the Alipay AI wallet is the world's first AI wallet aimed at individuals, supporting full visualization management of Agent tasks. Token Pay is the only comprehensive Token payment solution in the market, providing one-stop services to meet global market demands. AI Revenue offers global AI payment services for the AI industry, enabling 'pay-as-you-call' functionality, which includes features such as 'Payment MCP Server', 'AI Tipping' services, and 'Payment Integration Skills', providing multi-scenario capabilities that comprehensively cover leading development platforms and OpenClaw-like Agents. Alipay also stated that its 'AI Payment' has facilitated 300 million AI intelligent entity payments in the AI industry and supports 95% of general intelligent entities.

[Alipay Launches World's First Token Pay Service and AI Wallet Product, Showcasing Full-Stack AI Payment Solutions]According to monitoring by Dongcha Beating, Alipay has announced the launch of the world's first Token Pay service and AI wallet product. Today, the company's full-stack AI payment product matrix has been unveiled, marking the establishment of the world's first large-scale commercial AI-native payment infrastructure. Among these, the Alipay AI wallet is the world's first AI wallet aimed at individuals, supporting full visualization management of Agent tasks. Token Pay is the only comprehensive Token payment solution in the market, providing one-stop services to meet global market demands. AI Revenue offers global AI payment services for the AI industry, enabling 'pay-as-you-call' functionality, which includes features such as 'Payment MCP Server', 'AI Tipping' services, and 'Payment Integration Skills', providing multi-scenario capabilities that comprehensively cover leading development platforms and OpenClaw-like Agents. Alipay also stated that its 'AI Payment' has facilitated 300 million AI intelligent entity payments in the AI industry and supports 95% of general intelligent entities.

Alipay Launches World's First Token Pay Service and AI Wallet Product, Showcasing Full-Stack AI Payment Solutions

zest Surges 10.15% in 24H, Currently at 0.178793 USDT

【zest currently trades at 0.178793 USDT, up 10.15% in 24h, according to HTX market data.】zest Surges 10.15% in 24H, Currently at 0.178793 USDT

zest currently trades at 0.178793 USDT, up 10.15% in 24h, according to HTX market data.

Taiko, the layer 2 network of Ethereum based on zkRollup, released the Alpha-1 testnet (a1) Snæfellsjökull. The public testnet supports developers to deploy smart contracts, and opens up functions such as transaction testing and L2 node operation to ordinary users.

Detail

【Ethereum layer 2 network Taiko releases the first public testnet】Taiko, the layer 2 network of Ethereum based on zkRollup, released the Alpha-1 testnet (a1) Snæfellsjökull. The public testnet supports developers to deploy smart contracts, and opens up functions such as transaction testing and L2 node operation to ordinary users.

Ethereum layer 2 network Taiko releases the first public testnet

XRP Ledger Hard Fork In 8 Days? Upgrade Deadline Sparks Network Split Debate

As Presale Funding Nears $7.2 Million, Ozak AI’s Early Pricing Window Begins to Narrow

Market Expert Updates XRP Roadmap To $300 With New Data

The AI Stock Genius Who Made 60x Bets $7.7 Billion on Nvidia Topping Out

$50 Million Funding Ignites Airdrop Anticipation, Variational Becomes New Focus of Perp DEX

Musk Posted a Recruitment Ad for SpaceX, and After Reading the Comments Section, I Understood

Cloudflare CEO: How I Decided Which Employees to Replace with AI?

Secures Over $60 Million in Funding from Dragonfly, Sequoia, and Others. An In-Depth Look at the On-Chain Derivatives Protocol Variational | CryptoSeed

XRP Ledger Hits No. 4 In RWA Rankings In Less Than A Year

Warsh's First Day in Office, Markets Deliver a 'Wake-up Call': Rate Hike Expected This Year

Article

Trump Publicly Praises Micron, Speculation of Aggressive Call Option Purchases

Ethereum layer 2 network Taiko releases the first public testnet

Isenção de responsabilidade：O conteúdo acima não representa as posições da HTX.，A HTX não faz quaisquer recomendações de negociação.。

Artigos relacionados

Todos os comentários0Mais recentePopular

Tópicos em destaque

Artigos relacionados