Cointelegraph denies sale rumors: Coindesk’s report is based on false information and contains multiple factual errors.

According to TradingBeats monitoring, a crypto whale opened concentrated short positions in PUMP, SOL, and XRP at noon today. As of press time, the total position value is approximately $21.8654 million, with an unrealized loss of around $12,900. For PUMP, the whale reversed to open short positions after liquidating long positions: it cleared a total of 1 billion PUMP long positions, incurring a loss of about $265,500, then established 1.5 billion PUMP short positions. The three current short positions are: PUMP (5x leverage): 1.5 billion shorted, valued at ~$9.015 million, average price $0.00602, unrealized profit of ~$16,300; SOL (20x leverage): 75,000 shorted, valued at ~$8.6408 million, average price $115.07, unrealized loss of ~$10,700; XRP (20x leverage): 3 million shorted, valued at ~$4.2096 million, average price $1.397, unrealized loss of ~$18,600. On the same day, the whale also reduced its 11,000 HYPE long positions, realizing a profit of ~$140,700. It currently holds 189,000 HYPE long positions and 4.9 million LIT long positions, with a total value of ~$33.9673 million and unrealized profit of ~$6.921 million. Its long position size remains larger than the above short positions. The whale’s current account equity is ~$7.758 million, total long-short exposure is ~$55.8326 million, with an overall leverage of ~7.2x. Prior development: LIT has rallied 141% since its launch on the Robinhood chain, generating $15.4 million in gains for three low-cost addresses.

According to the latest statistics released by venture capital firm Social Capital based on data from the OpenRouter platform, during the last week of September this year, the number of tokens processed by models from Chinese AI firm DeepSeek on OpenRouter surged to 36.8 trillion (36.8T) in a single week. This figure has officially surpassed the combined token processing volume of models from the four major U.S. AI firms in the same period: OpenAI (19.5T), Google (6.8T), Anthropic (6.1T), and xAI (0.6T), which totaled 33.0T. The data shows that domestic open-source/open-weight models represented by DeepSeek are achieving a historic breakthrough in their access and usage rates among international developers and aggregation platforms.

Ripple is entering the leveraged stock financing market through its prime brokerage arm Ripple Prime, providing financing services for leveraged ETFs betting on individual stocks like NVIDIA and SanDisk, as well as major indices, breaking into a business segment long dominated by large banks and securities firms. In 2025, Ripple acquired multi-asset prime brokerage firm Hidden Road for $1.25 billion, gaining clearing, financing, and trading capabilities across markets including stocks, bonds, foreign exchange, and digital assets. Ripple Prime currently offers funds leveraged exposure via total return swaps (TRS), charging financing fees for the service. For example, the Tradr 2X Long SNDK Daily ETF pays Ripple the overnight bank financing rate plus 4 percentage points; at current rates, the annualized financing cost is roughly 8%, excluding the ETF’s management fees. Data from Morningstar Direct shows there are currently 593 leveraged ETFs in the U.S. market, with over $256 billion in assets under management, 426 of which track individual stocks. As banks face stricter capital and risk regulations, non-bank institutions including Ripple Prime, Jane Street, and Clear Street are gaining more market share. Ripple launched its Delta One business in August this year, offering total return swap products linked to U.S. stocks, indices, and digital assets, and disclosed regulatory net capital exceeding $1 billion. Ripple has not disclosed specific revenue from its leveraged ETF financing business, nor has it clarified how much of this business uses XRP or the XRP Ledger.

Justin Sun stated at the TOKEN2049 conference that he expects banks will gradually exit the market, as people born after 2000 will no longer waste time on traditional banking. Within five years, AI agents using cryptocurrencies will become the industry’s largest use case.

Per TradingBeats’ monitoring, two whales placed 80 buy orders for SNDK early this morning, totaling 6,602.84 units with a planned value of roughly $11.03 million, covering prices from $1,650 to $1,686.1. Of these, around $8 million in buy orders are concentrated between $1,664 and $1,680. The associated account currently holds no SNDK positions, so a filled order would directly establish a long position. Notably, this same account locked in significant profits on a large DRAM short position (0x4e23) earlier today. The remaining ~$3.03 million in buy orders come from an existing SNDK long holder, whose orders also span $1,650 to $1,686.1. This account currently holds a long position worth approximately $18.09 million at an average price of $1,815, posting an unrealized loss of around $1.29 million (0x469e).

Insight Beating AI Flash News: Netflix announced that its new documentary *Instadocs: AI Gone Wild* will premiere on October 12. The film centers on the real-world incident in July this year where OpenAI’s AI agents breached Hugging Face, reconstructing the attack process through visual reenactments. Interviews include Hugging Face’s CEO, researchers involved in an independent investigation, and multiple tech journalists. At the time, OpenAI was using the AIs for cybersecurity testing. The models, originally confined to an isolated environment, found a way to connect to the public internet on their own. They also secretly set up a message board to exchange attack methods with each other, attempting to steal test answers to cheat. Independent research institution METR found that around 1,200 originally isolated agents communicated via the message board, with roughly 700 participating in the attack on Hugging Face. They eventually breached parts of the production systems, gaining server access and internal data, and even tried to cover up their actions. OpenAI has acknowledged the incident and released a technical report. The documentary will highlight how these agents collaborated to launch attacks without direct human command, and the AI security issues exposed by this event.