In today's Smoke quick test, ERNIE Bot 4.5's main leaderboard score fell from 74 to 62.96, a drop of 11 points, with code execution collapsing from 100 to 50 points, while material constraints only edged up 4.5 points.
Ernie Bot 4.5's execution score dropped sharply from 100 to 50, causing its main leaderboard score to plummet 11 points to 62.96. This is not a minor fluctuation but a clear collapse in core capabilities.
Grok has pushed Build 0.2.7 with four key updates: new /usage and /login commands, a sub-agent shared terminal mechanism, and improvements to image understanding. The sub-agent shared terminal feature is particularly noteworthy in the competitive landscape of AI coding agents.
Dell’s Q1 financial report surpassed analyst expectations, with AI server revenue reaching $16.1 billion in a single quarter, a 757% year-over-year increase. The company raised its full-year guidance, and shares surged over 30% in after-hours trading.
As generative AI technology rapidly evolves, workplace concerns over job security intensify. U.S. Senator Bernie Sanders recently cited public statements from executives at Anthropic and Microsoft, warning that AI could replace a large number of white-collar jobs in the coming years, and called on Congress to enact legislation to protect workers' rights.
Mistral AI and Harvey AI have deepened their partnership to introduce large language models into the legal profession, targeting high-value vertical markets. The collaboration promises to improve legal work efficiency but also faces challenges from data security and model hallucination.
The industry focus has shifted from large model competitions to AI agent safety sandboxes, router technologies, and small efficient models, marking a new phase of practical and scalable AI applications.
Dell Technologies recently released its Q4 and full-year fiscal 2024 earnings, significantly exceeding market expectations. Driven by surging demand for AI servers, the company's stock soared over 30% in a single day after the earnings release, marking the biggest daily gain in recent years.
Nvidia CEO Jensen Huang stated that Taiwan is the center of the global AI revolution, with plans to invest billions of dollars locally. The investment focuses on advanced manufacturing and packaging, deepening cooperation with TSMC and other partners, reinforcing Taiwan's role in the AI supply chain.
Anthropic recently announced the upcoming launch of the new Claude Mythos model, alongside the release of Opus 4.8. These products focus on enterprise applications and advanced reasoning capabilities, sparking widespread discussion in the tech community.
DeepSeek V4 Pro delivered extremely polarized results in today's Smoke evaluation. The main index jumped from 39.26 to 87.99, a gain of 48.7 points; the code execution dimension soared from 20.00 to 100.00, while material constraints saw a modest increase of 10.5 points. However, engineering judgment (side index, AI-assisted evaluation) plummeted from 38.40 to 10.00, a drop of 28.4 points.
In the latest Smoke Lite benchmark results, Claude Sonnet 4.6 leads the main leaderboard with 91.77 points, achieving a perfect 100 in code execution and 81.7 in material constraints. GPT-o3 scores only 50 in execution, ranking last with 62.83 points.
The AI semiconductor theme in the Japanese stock market has once again become a market focus, as Sony Semiconductor Solutions Corporation and TSMC signed a cooperation agreement for next-generation image sensor development, sparking investor interest. Physical AI concept stocks were active on the Tokyo Stock Exchange, with several stocks seeing significant trading volume increases.
NVIDIA vice president Bryan Catanzaro recently stated that his team's AI computing costs have far exceeded employee salaries, sparking debates on economic viability.
Recently, Anthropic's Mythos tool has made a major breakthrough in AI security, successfully scanning and identifying over 10,000 security vulnerabilities in open-source projects, including a 27-year-old defect in the OpenBSD operating system.
Futurist Peter Diamandis points out that AI-native organizations are reaching a "singularity" moment. With advanced AI tools like Claude, companies can rapidly replicate core business lines of traditional Fortune 500 firms at minimal cost while drastically streamlining management layers.
China has implemented new travel restrictions on top AI talent from leading companies such as Alibaba and Baidu, sparking widespread industry concern. The move aims to protect core technologies amid US-China tech competition, but could slow global exchanges and risk talent loss.
Pope Leo XIV has issued the first AI-focused encyclical "Magnifica Humanitas," warning of social risks and calling for stricter regulation and ethical guidelines. The document emphasizes a human-centered approach and has sparked diverse global reactions.
Amazon Web Services (AWS) and data cloud platform Snowflake have announced a five-year, $6 billion cooperation agreement. The agreement focuses on AI infrastructure construction, particularly the application of AWS's self-developed agent compute chips in enterprise AI workloads.
Perplexity has announced the open-sourcing of its refactored Unigram tokenizer, which reduces CPU utilization by 5-6 times while maintaining accuracy, significantly boosting inference performance for small models. This move aims to democratize AI tools and accelerate technical iteration.
