Search Results: OpenAI

OpenAI has unveiled Agent Skills for Codex, enabling developers to extend the AI assistant with reusable, task-specific capabilities. This feature allows packaging instructions, scripts, and resources into modular skills for explicit or implicit invocation, streamlining workflows across CLI and IDE environments.

OpenAI’s latest GPT‑4.5 Turbo 2.0 pushes the limits of context length to 128 k tokens, slashes inference latency by roughly 40 %, and introduces a fine‑tuning API that scales to millions of examples. The update reshapes how developers build long‑form, context‑heavy applications and tightens safety controls across the ecosystem.

OpenAI has open-sourced a groundbreaking 0.4B-parameter sparse model from its 2025 research, enabling developers to probe neural circuits behind complex tasks like bracket counting and variable binding.

OpenAI introduces a comprehensive supply chain security initiative to address vulnerabilities in AI model development and deployment, aiming to standardize transparency and safety practices across the industry.

OpenAI’s announcement of GPT‑5 marks a pivotal moment for large‑language models, promising unprecedented accuracy, reduced hallucinations, and tighter alignment controls. The tech community is abuzz with implications for everything from enterprise productivity tools to policy frameworks that will shape the future of AI.

OpenAI has announced a sweeping policy shift that will no longer use copyrighted text to train its models. The move, aimed at addressing intellectual‑property concerns, could alter how AI systems learn and trigger a broader debate about data sourcing in the industry.