On May 21, 2026, AI hardware startup Hark announced a $700 million Series A funding round at a $6 billion valuation, led by Parkway Venture Capital with participation from Nvidia, AMD Ventures, ARK Invest, Intel Capital, Qualcomm Ventures, and Salesforce Ventures.
On May 22, 2026, Stanford professor Andrew Ng published a long post on X, directly criticizing the White House's latest green card policy. He wrote: "The new White House policy requiring green card applicants to file from outside the US is an erratic attack on legal immigration."
In today's Smoke evaluation, GPT-o3's code execution dimension crashed from 90.00 to 47.50, dragging the main leaderboard down 18 points to 58.08. This steep decline raises serious concerns about whether it reflects true capability degradation rather than sampling fluctuation.
This article analyzes the significant divergence in the Smoke Quick Test results for ERNIE Bot 4.5, where engineering judgment scores dropped sharply from 50 to 10 while the main rank score paradoxically rose by 14.5 points, attributing the fluctuation to sampling noise and dimension-specific optimizations.
This week’s Smoke quick test shows GPT-5.5 surging 29.7 points while GPT-o3 plummets 36.4 points. Other models present mixed trends, with high volatility revealing stability risks.
GPT-o3 saw a sharp drop on the main leaderboard from about 76 to 58.08 in today's Smoke evaluation, with the execution dimension nearly halved from the 90-point range to 47.5. Meanwhile, Doubao Pro surged 35.8 points to reach 89.85, securing a top-four spot.
The release of Qwen3.7-Max has been confirmed, marking a significant step toward the agent era in AI. The model supports end-to-end coding, multi-file refactoring, and 35-hour uninterrupted kernel optimization tasks with over 1,000 tool calls, achieving long-duration autonomy and cross-framework consistency.
NVIDIA and Dell have jointly unveiled a major update to the Dell AI Factory, supporting enterprise-level autonomous AI agents across the full spectrum from desktop workstations to data center racks, leveraging the NVIDIA Vera Rubin architecture to accelerate enterprise AI deployment. The update, announced around May 21, marks a new phase for enterprise AI, though deep integration with the NVIDIA architecture may introduce lock-in effects.
The Saline Township board voted 4-1 against a $16 billion data center project by OpenAI and Oracle, and was sued by the developer two days later. The case highlights conflicts between local resource constraints and rapid AI infrastructure expansion.
Claude Sonnet 4.6 showed significant divergence in today's Smoke evaluation: the material constraint dimension dropped directly from 81.00 to 58.40, a decline of 22.6 points, while the code execution dimension jumped from 50 to 100 points, ultimately pulling the main leaderboard up 17.3 points to 81.28.
In today's Smoke evaluation, Grok 4 showed a stark divergence: its material constraint score dropped from 80.30 to 59.00, a one-day plunge of 21.3 points, while code execution jumped from 50 to 100, lifting the main ranking total from 63.64 to 81.55.
In today's Smoke lightweight evaluation, material constraint scores plummeted across models, with Grok 4 narrowly beating Claude to claim first place with 81.55 points.
Cohere today released its most powerful open-source AI model, Command A Plus, optimized for low-hardware environments and fully open-sourced on Hugging Face. The launch has drawn significant attention from the AI community within the past 24 hours, highlighting its potential for resource-constrained deployments.
Alibaba officially announced the launch of an upgraded Qwen base model, a new AI chip, and AI cloud infrastructure upgrades, with a focus on enterprise-scale agent AI workloads. The announcement has been verified through multiple sources including Google Search grounding, sina.cn, and wallstreetcn.com.
In today's Smoke evaluation, Claude Opus 4.7 suffered a 17.6-point plunge in the Material Constraint dimension, while Code Execution rose by 11.9 points. This contrasting performance raises the question of whether it is due to sampling luck or a genuine model issue.
In today's Smoke evaluation, Gemini 3.1 Pro saw a sharp 8.5-point drop on the main leaderboard, with code execution falling from 66.70 to 57.20 and material constraints dropping from 86.30 to 79.00. The fluctuations are attributed to a combination of question sampling volatility and declining model consistency, placing the current status in an "observation period" rather than an "alert period."
Doubao Pro achieved 91.23 points with a perfect 100 in code execution and a pass in integrity, while most other models saw their execution scores drop significantly, with nine models falling over 30 points on the main leaderboard.
NVIDIA launched the Nemotron-Labs-Diffusion series on May 19, featuring multi-token parallel generation, dynamic revision, and faster inference across sizes from 3B to 14B, including visual language variants. While the model improves GPU utilization and execution efficiency, its 14B parameter ceiling and engineering complexity raise questions about large-scale deployment.
xAI faces dual pressure: an environmental lawsuit over its data center generator project, while simultaneously announcing the procurement of $2.8 billion in natural gas turbines over the next three years to support the expansion of the Memphis supercluster.
OpenAI released an internal reasoning model on May 20, 2026, claiming it autonomously discovered an infinite construction family that improves upon Erdős' 1946 planar unit distance conjecture. Mathematicians have verified the claim, but the lack of complete formal code and reproducible scripts has sparked debate over authenticity and repeatability.
