Anthropic recently made important adjustments to the usage limits for its Claude AI model, resetting the 5-hour and weekly limits for Pro and Max subscribers. This update mainly fixes the issue in Claude Code where excessive parallel generation of subagents rapidly consumed quotas, drawing widespread attention from the developer community.
xAI has officially announced the launch of its latest AI model, Composer 2.5, on the Grok Build platform. The model excels at handling long tasks and complex instructions, with significant improvements in response speed and intelligence, quickly sparking discussion in the developer community.
NVIDIA has announced that its Vera Rubin multi-rack system has officially entered full mass production, generating widespread attention in the tech industry. Designed specifically for agentic AI, the platform integrates multi-core components such as NVL72 to support global-scale intelligent agent deployment.
The state of Florida has filed a lawsuit against OpenAI and its CEO Sam Altman, alleging reckless misconduct in AI product development and deployment, and seeking legal accountability for violent incidents. This marks the first case globally to directly target corporate executives for AI safety responsibilities.
Anthropic has submitted a confidential IPO filing to the SEC with a valuation of $965 billion, surpassing market expectations and positioning it as a key competitor to OpenAI in the race to go public. Wall Street's enthusiasm for generative AI is accelerating the entry of these tech unicorns into capital markets.
The most direct finding from today's Smoke lightweight benchmark is that code execution ability no longer differentiates the top seven models, as all scored 100, making rankings entirely determined by material constraint scores.
SoftBank Group has announced a massive €75 billion plan to build Europe’s largest AI data center in France, a key move in the global AI infrastructure race to meet surging computing demand. The project, backed by French AI research and policy advantages, will roll out in phases and serve both global tech firms and local startups.
Alibaba Group has officially launched its latest large model Qwen 3.7-Max, priced aggressively at just one-sixth of Claude's comparable products, while delivering strong performance in multi-task processing and long-term stability. This move is seen as a major breakthrough for Chinese AI companies in the cost-effectiveness arena.
OpenAI recently announced that its advanced AI model has autonomously proven a geometry conjecture proposed by mathematician Paul Erdős over 80 years ago. The resulting 125-page proof has been recognized as rigorous by multiple mathematical experts, marking a milestone for AI in pure mathematics.
NVIDIA officially unveils its first chip designed for AI Agents, the RTX Spark, enabling local operation of large AI models on PCs without cloud dependency, marking a significant shift of AI computing toward edge devices.
OpenAI recently completed a $110 billion funding round, pushing its valuation to $730 billion. This news has sparked widespread attention in global tech and financial circles, setting a new record in the AI field and accompanied by rumors of a potential IPO by year-end.
This week, 270 translation tasks were completed by 3 models. Two samples were selected for multi-model blind evaluation, with the overall best being gpt-o3 (average score 9/10).
Grok's powerful image editing capabilities have sparked a viral trend on social media, with users creating and sharing AI-generated art, while raising discussions about ethics and privacy.
Anthropic engineers recently shared advanced workflow construction methods for the Claude model, sparking widespread attention in the AI community. This technical breakthrough focuses on "dynamic workflows and agent systems," replacing traditional manual prompt engineering with self-prompting mechanisms to significantly improve complex task processing efficiency.
OpenAI CEO Sam Altman recently announced on social platforms that the OpenAI Robotics team is actively recruiting hardware engineers, machine learning experts, and related talent, marking the AI giant’s official move into physical world robot applications. This initiative is seen as a strategic shift from pure software simulation to hardware-ML co-design.
Smoke's lightweight evaluation completed 10-question quick test on 11 mainstream models. Claude Sonnet 4.6 scored 99.78 points, while the GPT series collectively stuck at 74 points, highlighting the polarization in AI capabilities.
In a Harvard commencement speech, comedian Ronny Chieng urged graduates to "kill AI," drawing applause and igniting debate between supporters accusing him of anti-intellectualism and opponents seeing it as a signal of cultural shift.
WDCD Run #140 (2026-05-31) evaluated 11 frontier models on multi-turn commitment integrity, finding an average instruction decay of 36.5% from Round 1 to Round 3. Qwen3 Max led the field with 70.8 points and the lowest decay rate at 17%.
In the latest WDCD cycle, all 11 evaluated models show improvement in compliance ability, with the top five all rising and none declining. Gemini 3.1 Pro leaps into the top three with a +14.2 point gain, signaling a major shift in the competitive landscape.
The WDCD compliance test evaluates model stability under real enterprise constraints through three rounds of dialogue. The resource limitation scenario scored the lowest overall, becoming a common "stumbling block" for all 11 models.
