A British man has been sentenced to two and a half years in prison for his involvement in a SIM swap scam, and is charged with stealing $265,000 worth of cryptocurrency.

According to HTX market data, XDP surged more than 24% in a short period due to Bithumb’s launch of KRW trading pairs, and is now trading at $0.0243.

According to Nikkei Asia citing supply chain sources, Apple has requested some suppliers to cut parts orders for the iPhone 18 Pro and Pro Max, with October order volumes reduced by at least 15% from original plans, mainly due to rising memory chip costs and weaker-than-expected demand for the new models. This adjustment may be interpreted by the market as a reverse impact of expanding AI storage demand on the consumer electronics sector. AI data centers continue to compete for memory chip production capacity, driving up prices of DRAM and NAND. The starting price of Apple’s new high-end iPhone is $100 higher than the previous generation, which, while passing cost pressures to consumers, further suppresses end-user upgrade demand. However, the report notes that the order changes may also relate to Apple adjusting its iPhone new product launch rhythm, and cannot be directly equated with year-on-year declines in actual sales. Apple has not yet officially confirmed the relevant order-cut news. Affected by the news, Apple’s U.S. stock fell nearly 2% in pre-market trading. According to TradingBeats monitoring, the AAPL perpetual contract on Hyperliquid has fallen back to around $334.

According to an official announcement, Bithumb will list the XDP/KRW trading pair.

Beating AI Insight News: OpenRouter has developed an AI agent called Rasp for its 5-person sales team, dedicated to handling administrative tasks in the sales workflow. When a client sends an inquiry, Rasp researches their background, qualifies leads, and responds to emails. It prepares materials ahead of meetings and organizes minutes, updating client information post-meeting. Currently, 93% of initial outreach emails are sent automatically by the AI, with many clients receiving replies in under a minute. The time spent on pre- and post-product demo tasks has shrunk from 103 minutes to 44 minutes per session. OpenRouter notes the system saves approximately 600 hours monthly, with a daily cost of just $30, including engineering labor and model fees. Sales teams remain responsible for meetings, negotiations, and final decision-making. Rather than letting AI directly close sales, delegating repetitive tasks such as research, note-taking, and form-filling to AI first may yield better results.

CryptoQuant analyst MAC_D published a note stating that Bitcoin has pulled back roughly 7.61% from its October 5 high of $87,027 to its October 9 low of $80,404. This pullback is close to the 7.58% average pullback seen in the 2022–2025 bull cycle, but remains lower than the overall historical average pullback of 14.39% he compiled. Using Bitcoin bull cycle pullback metrics, MAC_D compared the declines from local highs across each bull run. The average pullbacks for the 2010–2011, 2011–2013, 2015–2017, and 2018–2021 cycles stand at 18.41%, 21.71%, 10.93%, and 19.39% respectively, marking the pullback in the most recent cycle as relatively mild. He pointed out that applying the overall historical average pullback of 14.39% to the recent high would result in a corresponding price of roughly $74,500, which can serve as a reference for further downside. MAC_D suggested investors consider accumulating BTC gradually around $80,000 and increasing their buying volume in the $74,000–$75,000 range.

Beating AI News Brief: Intelligent Internet, the new venture of Stability AI co-founder Emad Mostaque, has open-sourced Evoke, a PostgreSQL database search plugin with a built-in AI model of roughly 30.3 million parameters. Once installed, developers can directly search documents and knowledge bases within the database, supporting both keyword matching and semantic similarity searches. Built on IBM’s Granite-Embedding-30M-Sparse model, Evoke generates a set of weighted relevant terms for documents, enabling retrieval of matching content even if the search query does not appear in the original text. Keyword and semantic information share a single search index, eliminating the need for separate vector index maintenance. The model runs natively on standard CPUs, requires no GPU, and does not involve sending data to external AI services. Official English retrieval tests previously released show Evoke achieves a 66.7% relevant document retrieval rate in its top 100 results, outperforming traditional keyword search (56.3%) and nearly matching the 67.1% rate of a 600-million-parameter model. Evoke’s code and model are both open-sourced under the Apache-2.0 license, making it suitable for use cases including internal enterprise document retrieval and AI Agent knowledge bases.