| Mobile Web

Search results for GPT

AI & Enterprise

Three reasons DeepSeek\'s new V4 model matters: high performance, low cost and Chinese AI chips

Chinese AI company DeepSeek unveiled its next flagship model, V4, stepping up its push into the open-source AI market. The model expands long-context handling, improves price competitiveness and highlights optimisation for Chinese-made chips. DeepSeek released two versions, V4-Pro and V4-Flash, with web, app and API access. It also touted benchmark performance and resource reductions versus its prior model. V4 is its first model optimised for Huawei Ascend chips, though analysts say a full shift away from Nvidia is unlikely yet.

AI & Enterprise

Nvidia supplies GPT-5.5 Codex to 10,000 employees, cutting cost to one-35th

OpenAI’s GPT-5.5-based Codex has been deployed on Nvidia Blackwell systems, with more than 10,000 Nvidia employees beginning to use it. Nvidia said the rollout cut costs to one-35th of GPT-4o levels and increased token output per megawatt by 50 times. Codex has expanded beyond coding help to an agent tool supporting non-technical work across departments. Nvidia employees praised the results, and Jensen Huang said agents do work beyond answering questions.

AI & Enterprise

U.S., China step up next-generation LLM race as big tech reshapes coding AI

Competition in large language models is intensifying as leading AI companies in the United States and China release new systems more frequently. OpenAI unveiled GPT-5.5 and began distributing it to paying ChatGPT users, while Anthropic launched Claude Opus 4.7. China’s DeepSeek released preview versions of V4 and Moonshot AI unveiled Kimi-K2.6. The coding AI market is also being reshaped by capital-heavy companies, including SpaceX securing an option to acquire Cursor for $60 billion.