Hong Kong's Hang Seng Index closed up 1%, with Zhipu surging 7.59%.

Insight Beating AI Flash Briefing: Large language models can now learn language independently, but how text is split into tokens is usually determined by pre-trained tokenizers. This saves computing power, yet also constrains how models structure information. Fixed segmentation rules often fall short for unfamiliar words or code; skipping tokenization entirely to process raw bytes directly, however, leads to a sharp rise in computing costs. AI voice firm Cartesia introduced H-Net (Hierarchical Network) in July 2025 to resolve this tradeoff. The framework lets models read raw bytes directly and learn to segment and compress information autonomously. For example, an unfamiliar word may initially split into multiple segments; as it appears repeatedly in context, it gradually merges into fewer segments. This adjustment can occur during inference, with no retraining required. Cartesia has now unveiled a new H-Net version, with notable gains in training efficiency. In experiments with strictly controlled key conditions, computing power needed to hit the same validation loss dropped by ~22%, and computational efficiency for downstream evaluations nearly doubled. In larger-scale trials, the new H-Net matched the average score of open-source model SmolLM3 (trained on ~11 trillion tokens) across six tests using only ~1 trillion token-equivalent data, requiring roughly one-tenth the training compute of SmolLM3. This approach is not limited to text: the team trained H-Net on bacterial and archaeal genomes, finding that when processing human gene fragments, the model could automatically recognize the pattern of grouping every three bases in coding regions. Related DNA modeling experiments saw a 6–7x boost in computational efficiency. Unlike text, speech has no pre-defined word boundaries. Cartesia’s initial H-Net research was actually aimed at enabling voice models to learn phonemes and words directly from continuous audio, without relying on pre-designed segmentation rules.

According to HTX market data, HYPE continues its rally, surging over 7% in the past hour and currently trading at $85.2840. Earlier reports noted that Intercontinental Exchange (ICE), parent company of the New York Stock Exchange, plans to collaborate with Trade.xyz to introduce perpetual contract products into regulated U.S. markets.

According to TradingBeats monitoring, over the past two hours, a whale wallet starting with 0x30a opened a new long position in CL crude oil worth approximately $17.9554 million at an average price of $100.9, using 10x isolated leverage. The position currently has an unrealized profit of around $69,300, with a liquidation price of ~$93.19, roughly 8.07% below the current market price. The account’s equity stands at about $2.8723 million; its combined long positions in crude oil and Microsoft total around $28.846 million, with an overall leverage ratio of roughly 10.04x. Crude oil accounts for 62.45% of total holdings, while the remaining portion is a Microsoft long position worth approximately $10.832 million. Fund records show that between October 26 and 27, the account received two transfers totaling ~2.0312 million USDC from the same address. The profit curve indicates its overall net loss over the past month is around $14.9545 million. Before shifting to crude oil, the whale took profits on its long positions in Western Digital and Intel this morning after their earnings reports, generating a total trading profit of about $714,200; additionally, part of its Google long position was liquidated, leading to a total net loss of roughly $590,000 upon exiting those positions.

Bitunix has launched customer service via its official account @Bitunix_Support on X, handling all user needs—including account issues and platform usage inquiries—around the clock. Users can access support conveniently without leaving the X platform, with timely responses available even outside standard working hours. This development further enriches Bitunix’s customer service channels, underscoring the platform’s commitment to continuously enhancing service accessibility and user experience, as it aims to deliver more attentive, efficient support to users worldwide. Official customer service account: https://x.com/Bitunix_Support

According to Bloomberg, Intercontinental Exchange (ICE), the parent company of the New York Stock Exchange, is collaborating with Trade.xyz to introduce perpetual contract products to U.S. regulated markets. The market will feature independent clearing, know-your-customer (KYC) and anti-money laundering (AML) systems. It will be built on Hyperliquid’s platform, utilize Trade.xyz’s products linked to stock indices, commodities and currency prices, and is currently awaiting regulatory approval.

According to TradingBeats monitoring, BTC is trading at around $80,131. Two whales hold a combined short position of approximately 331 BTC, valued at about $26.523 million, with liquidation prices concentrated between $80,861 and $80,867 — just 0.91% to 0.92% away from the current price. Among them, the short position holder (0xa4de) with a $16.026 million position has an average entry price of $79,913.7, an unrealized loss of around $43,400, and a liquidation price of $80,861. The other short position holder (0x92a7) with a $10.497 million position has an average entry price of $79,879.9, an unrealized loss of about $32,900, and a liquidation price of $80,866. The first whale has set a full-position market stop-loss trigger order at $80,865, with the trigger price roughly $3.71 higher than the liquidation price, and is expected to execute a full stop-loss after the first round of liquidation is completed.