GPGraphite ราคา

dna Surges 14.34% in 24H, Currently at 0.1969 USDT

On June 12, Hubei Jiangcheng Laboratory announced a significant breakthrough in capacitor technology, successfully developing a 3D multilayer on-chip capacitor with a capacitance density exceeding 1000 nanofarads per square millimeter. This capacitor can be directly applied to high-end chips such as AI/GPU chips and high-performance processors, supporting the development of high computing power and low power consumption chips. Currently, related technologies are undergoing process flow testing and small batch pilot production, aiming for large-scale application in advanced packaging. In circuits, capacitors essentially act as ultra-miniature 'reservoirs': they can quickly charge and discharge to stabilize voltage when chip current fluctuates dramatically, ensuring that the chip receives 'pure and stable' current. In computing power systems, capacitors are also likened to 'electric RAM': HBM serves as data buffering for computing power, while capacitors act as energy buffers. To supply power instantly to GPUs at full capacity, a multi-level power caching system from nanoseconds to seconds is essential.

According to monitoring by Beating, Romain Huet, the head of developer experience at OpenAI, revealed in a response to user inquiries on X that starting from GPT-5.4, the Codex model line (a series of programming-specific models previously maintained independently by OpenAI, distinct from the similarly named desktop application) has been merged into the main model, and will no longer be iterated separately. GPT-5.5 further enhances this integration, showing significant improvements in areas such as agentic coding (agent-driven autonomous programming) and computer use (manipulating computers to complete tasks). The practical impact of this merger for developers is that there is no longer a need to choose between a general model and a programming model; one model now encompasses dialogue, reasoning, and programming capabilities.