Крупные игроки провели серию значимых транзакций

cryptonews.ruPubblicato 2021-06-21Pubblicato ultima volta 2024-08-21

В последние минуты на криптовалютном рынке зафиксировани значительные перемещения крупных объемов цифровых активов, что мгновенно привлекло внимание аналитиков и инвесторов. Исследователи Whale Alert отметили сразу несколько крупных транзакций, включая перевод 12 099 ETH на сумму около $31,9 млн с неизвестного кошелька на адрес компании Cumberland, известной своим участием в криптовалютных операциях.

По мнению аналитиков, это может указывать на подготовку к крупной сделке или реорганизацию активов, что заслуживает пристального внимания. Другие значимые транзакции включают перевод 13 745 ETH около $36,2 млн с биржи Kraken на неизвестный кошелек. Этот шаг может свидетельствовать о выводе средств для долгосрочного хранения или участие в внебиржевых сделках, что является распространенной практикой среди институциональных инвесторов, стремящихся минимизировать влияние на рынок.

Еще одна значительная операция — перевод $50 млн в USDT с казначейства Tether на неизвестный кошелек. Такие переводы часто указывают на предстоящие значительные покупки или поддержку ликвидности на биржах. Однако остается вопрос, будет ли этот капитал использован для приобретения криптоактивов или задействован в торговых стратегиях.

Стоит отметить и обратный перевод 12 115 ETH на сумму около $32 млн с кошелька Cumberland на неизвестный адрес, что может указывать на распределение активов между различными счетами для диверсификации рисков или проведения дальнейших операций.

Эти крупные движения средств подчеркивают текущую активность институциональных игроков на криптовалютном рынке, особенно в условиях повышенной волатильности и неопределенности. Инвесторам следует внимательно следить за дальнейшими перемещениями, так как они могут служить индикаторами будущих ценовых изменений или значительных перемен в стратегии крупных игроков.

Letture associate

Cache hit rate reaches 99.93%, the best Harness for DeepSeek is here, GitHub pulls in 86k Stars

The open-source programming agent "Pi" has surged in popularity on GitHub, amassing around 86,000 stars, largely due to its exceptional efficiency when paired with DeepSeek's models. A key highlight is its remarkably high cache hit rate of approximately 99.93% for input tokens when using DeepSeek. This minimal cache miss rate of 0.07% means most repeated context is reused, drastically reducing computational costs. In a benchmark test by Composio comparing eight major agent harnesses running DeepSeek V4 Flash, Pi achieved the lowest average cost per successful task at about $0.028, significantly outperforming others like Claude Code, which was nearly seven times more expensive. Pi functions as a streamlined "harness" or workspace for AI models, providing core tools like file read/write and command execution while allowing users to add extensions. Its simplicity and specific adaptation for DeepSeek's API—handling its unique reasoning content and tool call formats—make it highly compatible. This design maximizes the efficiency of DeepSeek's low-cost tokens and powerful caching mechanism, as the agent's linear workflow avoids unnecessary changes to earlier prompt sections. While DeepSeek has officially announced its own harness project targeting Claude Code, Pi remains a versatile, user-customizable platform supporting multiple AI providers. It offers developers a cost-effective and flexible alternative, setting a high bar for future official solutions.

marsbit7 min fa

Cache hit rate reaches 99.93%, the best Harness for DeepSeek is here, GitHub pulls in 86k Stars

marsbit7 min fa

EthSystems and Undefined Labs sign MOU to support institutional Ethereum adoption in South Korea

EthSystems, an institutional Ethereum infrastructure firm founded by the former leaders of the Ethereum Foundation's Institutional Privacy Task Force, has signed a strategic Memorandum of Understanding (MOU) with South Korean Web3 advisory firm Undefined Labs. The partnership aims to accelerate institutional Ethereum adoption in South Korea by collaborating with financial institutions, regulators, and enterprises. Key focus areas include tokenization, stablecoins, digital asset infrastructure, and privacy-preserving blockchain systems. EthSystems, which spun out from the Ethereum Foundation in July 2026, provides technical advisory and infrastructure to help institutions deploy compliant, production-ready financial applications on Ethereum. Undefined Labs will support market entry, client introductions, business development, and provide local regulatory insights. This collaboration combines EthSystems' institutional Ethereum expertise with Undefined Labs' deep knowledge of the Korean financial market. Together, they will support Korean institutions through education, workshops, proof-of-concept initiatives, and long-term ecosystem development as Ethereum becomes a foundational element for next-generation financial infrastructure.

cointelegraph12 min fa

EthSystems and Undefined Labs sign MOU to support institutional Ethereum adoption in South Korea

cointelegraph12 min fa

When Meme Traffic Meets RWA: The Financial Ambitions of Hyperliquid and Robinhood

When Meme traffic meets RWA: Hyperliquid and Robinhood's Financial Ambitions This article explores the transformative role of speculation in financial market evolution, arguing it can provide the liquidity and capital foundation for mature, compliant businesses. The analysis draws parallels to historical examples, like 19th-century Chicago grain markets, where speculators provided necessary counterparty liquidity for farmers, eventually establishing a global pricing infrastructure. The author examines contemporary cases in crypto. The launch of Uniswap's Pools meme token platform on Robinhood Chain—originally built for tokenized stock trading—demonstrates how speculative meme trading can drive significant user adoption and liquidity. Despite initial volatility, this traffic can seed future, more substantial financial activities. Similarly, Hyperliquid, which began with high-leverage crypto trading, has successfully migrated its speculative model to real-world assets (RWAs) like equities, commodities, and indices. By July, RWA perpetual contract volumes on Hyperliquid surpassed its crypto trading for the first time, showcasing how speculation can build new, all-weather pricing layers. The core thesis is that speculation itself is neutral. Its long-term value and sustainability are determined by the underlying assets it engages with. While pure bubbles collapse, speculation anchored to valuable assets—be it wheat, gold, or equities—can evolve into foundational market infrastructure. Both Hyperliquid’s RWA platform and Robinhood’s strategy of leveraging meme coin enthusiasm to build towards on-chain securities trading exemplify this potential pathway from speculative frenzy to structured financial utility.

marsbit31 min fa

When Meme Traffic Meets RWA: The Financial Ambitions of Hyperliquid and Robinhood

marsbit31 min fa

NVIDIA HBM in Short Supply, Next-Gen GPUs Limited, but This Storage Drive Steals the Spotlight

The article discusses how AI Agents are transforming storage from a passive repository into an active component within AI inference and operation loops. As Agents perform continuous tasks involving models, memory, tools, and logs, data storage needs to evolve beyond simple block devices. The concept of "functional SSDs" is introduced, where capabilities like automatic encryption, compression, indexing, and memory management are embedded closer to the storage medium. This shift is driven by the need to handle Agent-specific data chains—including context, tool trajectories, and long-term memory—more efficiently. The piece analyzes current trends like AI SSDs from companies such as Phison (aiDAPTIV), Longsys (SPU+iSA), and Maxio, which are beginning to participate in the AI data path by managing model weights, KV cache, and prefetching. It further explores the future re-division of labor across the memory hierarchy: HBM for core compute, emerging High Bandwidth Flash (HBF) for read-intensive workloads, DRAM/CXL for mutable state, and functional SSDs for persistent, governed objects like Agent Memory. The conclusion is that storage will become integral to Agent capability, moving from just saving data to enabling next-step actions. The industry is poised to develop along three paths: functionalized SSDs, storage nodes tailored for Agents, and a re-architected, tiered memory system optimized for access patterns, security, and cost-per-token efficiency.

marsbit56 min fa

NVIDIA HBM in Short Supply, Next-Gen GPUs Limited, but This Storage Drive Steals the Spotlight

marsbit56 min fa

Breaking: Claude's Attempt on Riemann Hypothesis "Fails," but Accidentally Sets 37-Year Math Record

**AI Makes a Major Mathematical Leap While Failing at the Riemann Hypothesis** A research version of Anthropic's Claude AI was tasked with "seriously attempting" the legendary Riemann Hypothesis, a 167-year-old unsolved problem in mathematics. While it ultimately did not prove the hypothesis, its attempt yielded a significant breakthrough. Claude managed to improve the proven lower bound for the proportion of Riemann zeta function zeros lying on the critical line from 41.6% to 67.2%. This marks a 25.6 percentage point increase, described by observers as potentially the most substantial advance in analytic number theory in over a decade, considering that the previous record had only improved by 0.8 percentage points over 37 years. To achieve this, Claude generated and discarded roughly 650 initial ideas before orchestrating a day-and-a-half-long effort involving about 60 sub-agents. These agents executed 2,400 shell commands, wrote hundreds of Python scripts, and performed thousands of numerical checks. After finding the potential result, Claude initiated self-verification, downloaded papers to check for prior work, and independently re-derived the finding. It then suggested writing a formal paper. The result has been reviewed by Anthropic's internal mathematicians, who also collaborated with Claude to produce a formal, machine-checkable proof using the Lean theorem prover. External experts in the field have also reviewed the manuscript. Anthropic clarifies that the method likely does not directly lead to a full proof of the Riemann Hypothesis, but it demonstrates that advanced AI models are beginning to engage with genuine, open-ended research problems at the frontier of mathematical knowledge.

marsbit1 h fa

Breaking: Claude's Attempt on Riemann Hypothesis "Fails," but Accidentally Sets 37-Year Math Record

marsbit1 h fa

Trading

Spot
活动图片