Lighter Officially Launches Its EVM Development Network

According to data from mempool.space, Bitcoin completed its latest mining difficulty adjustment at block height 971712 at 20:37:31 Beijing Time today, with the difficulty rising by 5.92% to 140.58T.

According to market data from BIT (bit.com), Strategy (MSTR) has extended its decline to 6.46%, currently trading at $141.5. On the news front, MSCI plans to adopt new rules that screen out companies whose asset structures and operational characteristics are closer to investment funds than traditional operating enterprises, based on tests of operating asset ratios and multiple financial metrics. Strategy, due to its extremely high proportion of Bitcoin assets, has been included in the watchlist of potential removals. The final affected list is expected to be announced on November 11. Market participants believe that if MSTR is excluded from MSCI-related indices, passive funds tracking these indices will be forced to reduce their holdings of the stock, potentially creating billions of dollars in selling pressure and posing a challenge to its model of continuously increasing Bitcoin holdings through equity financing.

Insight Beating AI Flash News: Liquid AI has released the multimodal decision model d1-ultra, which scored 65 points in Artificial Analysis’s Decision Index v1.0, surpassing Jev 1.14’s 57 points to claim the top spot. Unlike models built for chatting or article generation, this decision model reads text and images to directly judge which option is more likely correct and provides corresponding probabilities. Its applications include inspecting circuit board defects, verifying that product images match their descriptions, or reviewing customer-uploaded claim photos. However, the new model’s top ranking comes with a notable tradeoff: in Artificial Analysis’s 500-question knowledge test, d1-ultra answered only ~32% of questions correctly, lower than the previous d1 generation’s 41%. Its strength lies in reducing incorrect responses. Among questions it did not answer correctly, the share of choosing “I don’t know” rose from 2% to 85%. When facing uncertain queries, it is more willing to admit ignorance rather than risk providing wrong answers — a key factor behind its significant ranking improvement. Artificial Analysis assigns 12.5% weight each to knowledge accuracy and the ability to avoid incorrect responses, with the remaining score derived from complex reasoning, tool call judgment, and safety classification. Liquid AI’s own 112-test evaluation shows d1-ultra outperforms Jev 1.14 across all eight task categories. d1-ultra is available via Liquid API and Vercel; its model weights have not been open-sourced. It charges $0.042 per million input tokens, 2.1 times the price of the previous d1 model, with no output fees. For a single image used to answer multiple questions, input costs are recalculated for each question.

According to market data from BIT (bit.com), the three major U.S. stock indexes showed mixed performance: the Nasdaq Composite fell 1.09%, the Dow Jones Industrial Average rose 0.44%, and the S&P 500 slipped 0.11%. AI-related stocks declined broadly, with individual components posting losses as follows: COHR down 5.89%, LITE down 4.06%, AAOI down 5.83%, SKHY down 5.13%, STX down 5.84%, WDC down 4.84%, SNDK down 4.29%, and MU down 4%.

Beating AI News Flash: China’s Ministry of Commerce and Ministry of Science and Technology plan to include 23 AI technologies in export controls, covering large language models (LLMs), AI agents, computing power clusters, and more. For LLMs, core technologies may be restricted if they meet any of three criteria: 1. Training compute reaches 10^26 floating-point operations; 2. Parameter count hits 2 trillion; 3. Ranks in the top 10 of global mainstream public evaluations. The scope of controls includes model weights, underlying source code, architecture design, and training optimization technologies. However, this does not mean Chinese LLMs cannot be open-sourced going forward. Open-source technologies already in the public domain are not bound by the new rules, and future open-sourcing of restricted technologies can go through a fast-track licensing process. The Ministry of Commerce stated that this process can basically achieve "immediate approval and export". Overseas users continuing to use Chinese models via APIs, web interfaces, or apps also fall outside the scope of this LLM technology export control. What is truly restricted is the transfer of core technologies abroad, such as handing over model weights to overseas enterprises or conducting localized deployment overseas. Some technologies involving China’s core data, national defense, and other sensitive fields are proposed to be banned from export. The new catalog is currently open for public consultation, with a deadline of November 15, while the pre-export registration requirement for related technologies has already taken effect on October 16.

Insight Beating AI Flash News: U.S. monitoring software firm Datadog has launched its first dedicated operations and maintenance (O&M) model, Scout 1.0, built on Zhipu AI’s open-source GLM-5.3 Flash. Designed specifically for troubleshooting online system failures, Scout can query logs, monitoring metrics, and service call traces to gradually identify root causes of issues. Datadog built a simulation environment using real incident and monitoring data, then trained the model via reinforcement learning to eliminate false clues and support diagnostic conclusions with evidence. In 90 fault-troubleshooting tasks from public benchmark SREGym, Datadog’s self-test of Scout’s highest inference tier completed 80 diagnostic and repair tasks, tying with GPT-6 Astra xhigh and only 2 tasks fewer than GPT-6.1 Sol xhigh. The original GLM-5.3 Flash finished 66 tasks. In Datadog’s internal AlertEval evaluation, Scout’s mid-tier diagnostic quality is close to GPT-6.1 Sol high and Claude Opus 5.5 high, with an estimated inference cost 3 to 8 times lower. Compared to GPT-6 Astra medium, the cost is roughly 19 times lower. Scout has been integrated into Datadog’s Bits Investigation and is currently available in a closed preview to a small number of customers. At this stage, it mainly focuses on identifying fault causes and proposing recovery suggestions; automatic repair remains a future R&D goal.