Tencent's Hunyuan launches real-time voice model Hy Realtime: it can speak, sing, and understand impromptu revisions.

Insight Beating AI Flash News: Mirendil, founded by a former Anthropic researcher, has closed a $1 billion Series A funding round led by Kleiner Perkins and Clearlake Capital, with participation from Google, Battery Ventures, and other investors. Less than five months after its $200 million seed round in June this year, the company’s total funding has reached $1.2 billion. The Wall Street Journal, citing sources familiar with the matter, reported that the current round values Mirendil at $5 billion, five times its seed round valuation, though the company has not officially disclosed its valuation. Mirendil’s goal is to use AI agents to handle all R&D work for an AI lab. Last month, it unveiled an experiment: a researcher, working with AI agents, pre-trained Mirendil-Base from scratch in just three and a half months, with over 99% of the code written by agents and a compute cost of less than $10 million. Shortly after, the company launched the Fern model, which supports a 10 million token context window. At the end of October, Mirendil also rolled out Lor, a multi-agent R&D platform designed to let AI agents collaborate on model development. The platform can schedule hundreds of agents simultaneously to write code, run experiments, automatically allocate and release GPUs, and operate continuously for days or even weeks. Mirendil also tested Lor’s capability to schedule 102 agents at once to set up test environments for 100 PyTorch code modifications, with an average of 8 minutes per task. Mirendil aims to make this technology accessible to external research teams, allowing scientists in fields like pharmaceuticals, materials science, and robotics to develop specialized models without building large in-house AI teams. In addition to supporting its own Fern model, Lor can integrate models from Claude, GPT, Kimi, DeepSeek, and others. The platform is currently in an invitation-only beta testing phase, with free access during the test period.

Renowned DeFi researcher Ignas noted in a post that for assets like Bitcoin, which rely on sentiment and consensus for support, price is the product — the lower the price, the weaker the market’s belief in its "store of value" narrative. This cycle’s downturn has been particularly painful; many had firmly believed $58,000 was the bottom, but BTC’s failure to rally alongside AI stocks dealt a major blow to market confidence. As a result, the current slump lacks meaningful bottom-fishing buying, with capital instead flowing toward AI assets. Ignas emphasized that if BTC continues to plunge, there may still be buying for bounce plays, but these are likely short-term speculators rather than long-term holders. Once, during downturns, crypto Twitter was flooded with bulls advocating for sound money theories; now, it’s eerily quiet. More worrying is MSCI’s delisting of Strategy; the lower BTC falls, the stronger the negative panic surrounding Strategy grows. The most bizarre development is the performance of altcoins: they have remained surprisingly resilient relative to BTC, with Bitcoin instead dragging the entire market down.

，散户投资者在 11 月 10 日净买入约 7.3 亿美元英伟达股票，为 3 月以来单日最大买入规模。此前 11 月 9 日净买入约 5.9 亿美元，11 月 11 日再净买入约 3 亿美元。三个交易日合计净买入约 16 亿美元，为至少 12 个月来最大的三日买入。同期，2 倍做多英伟达 ETF NVDL 录得约 4.01 亿美元流入，有望创下 2025 年 11 月以来最大单周流入。另一只 2 倍做多英伟达的 ETF NVDU 录得约 6400 万美元流入，有望创下 2026 年 2 月以来最大单周流入。11 月 12 日开盘后的前 30 分钟，散户买入约 4.55 亿美元、卖出约 3.45 亿美元，净买入约 1.11 亿美元，为有记录以来开盘前 30 分钟的最大净买入，散户正在加码买入英伟达。

Dongcha Beating AI Flash News: Figure AI founder Brett Adcock’s personal AI assistant Hark has launched a new-generation web operation model Handoff2, which is now available to all Hark Pro users. Compared with the previous version, the new model doubles the average overall speed of completing web tasks, with flight booking tasks up to 3.8x faster, and long research tasks even over 5x faster. Hark tested the new model on 478 tasks across 238 websites, and its task completion rate rose from 73% to 82% (a 9-percentage-point increase). Its CAPTCHA handling capability also improved by 80%. According to Hark’s self-test, Handoff2 outperforms Claude Opus 5.5 High in both web operation performance and speed. Compared with the higher-tier Opus 5.5 Max, the accuracy gap is less than 5 points, but its task completion speed is about twice that of Opus 5.5 Max. Hark mainly improved speed by optimizing reinforcement learning training: during training, it penalizes excessive reasoning, repeated operations, and time-consuming steps, teaching the AI to execute multiple operations at once, such as filling multiple form fields consecutively or reusing repeated action combinations. After training, the single-task time of the new model was reduced by 46%, coupled with backend optimizations, leading to an average overall speed twice that of the previous generation. Hark also disclosed that since its official launch in October this year, Handoff has run more than 5.6 million browser sessions and processed over $20 million in shopping transactions.

Beating AI Express (Dongcha) reports: Anthropic’s official status page shows rising error rates for claude.ai, Claude Code, and Claude Console (platform.claude.com). The company is investigating login issues affecting users of Claude.ai and Claude Code, and will provide updates promptly. Currently, claude.ai is marked as a major outage, while Claude Console and Claude Code are partial outages. The outage also impacts Claude Cowork, a service that had previously been suffering from performance degradation. Claude API and Claude services on cloud platforms including AWS, Google Cloud, and Microsoft Foundry remain operational.

JPMorgan’s latest 13F filing shows that as of September 30, it held 848,900 common shares of Strategy, with a market value of approximately $131.3 million calculated at the end-of-quarter price. Compared to the 624,400 shares held at the end of the second quarter, the share count increased by 224,500, a rise of around 36%. During the same period, the firm held 11,408,900 shares of BlackRock’s Bitcoin ETF (IBIT), with a market value of about $540 million, up approximately 9.6% from the 10,407,600 shares held at the end of Q2. The same filing also lists 1,581,800 common shares of Coinbase, valued at around $301 million, as well as stakes in Robinhood and listed Bitcoin mining companies including Core Scientific, MARA Holdings, Riot Platforms, Hut 8, and CleanSpark. The filing further shows JPMorgan holds a large number of Strategy put and call options: the put options correspond to around 1,148,500 shares with a notional value of approximately $178 million, while the call options correspond to around 349,500 shares with a notional value of $54.06 million.