Ce que OpenAI a publié en mars 2026

OpenAI acquires TBPN

OpenAI has acquired TBPN, a daily live tech talk show and media company focused on AI, tech, business, and culture. The acquisition aims to enhance OpenAI’s communication strategy by leveraging TBPN’s…

Accelerating the next phase of AI

OpenAI has secured $122 billion in new funding, marking one of the largest capital raises in AI history. This funding will accelerate global expansion of frontier AI technologies, including investment…

Pricing discovered for Openai

Pricing updated for Openai: - New tier: Free (Custom pricing) - New tier: Go (Custom pricing) - New tier: Plus ($20/mo) - New tier: Pro ($200/mo) - New tier: Business (Custom pricing) - New tier: Ente…

Helping disaster response teams turn AI into action across Asia

STADLER reshapes knowledge work at a 230-year-old company

OpenAI highlights how Stadler, a global rail vehicle manufacturer, has implemented ChatGPT to enhance knowledge work efficiency across its 650 employees. The initiative focuses on leveraging AI to str…

How we monitor internal coding agents for misalignment

OpenAI has developed and deployed an internal monitoring system to detect misaligned behavior in its coding agents, leveraging its most advanced models, including GPT-5.4 Thinking at maximum reasoning…

Inside our approach to the Model Spec

OpenAI introduces the Model Spec as a public framework to guide the behavior of its AI models, emphasizing a balance between safety, user freedom, and accountability amid advancing AI systems. This fr…

Introducing GPT‑Rosalind for life sciences research

OpenAI introduced GPT‑Rosalind, a frontier reasoning model tailored for life sciences research, including biology, drug discovery, and translational medicine. The model is optimized for scientific wor…

Creating with Sora safely

OpenAI has introduced Sora 2, a new video generation model, alongside the Sora app, emphasizing built-in safety as a core feature. The announcement highlights a commitment to integrating safety measur…

Our response to the Axios developer tool compromise

OpenAI disclosed a security incident involving a third-party developer tool, Axios, which was compromised as part of a broader software supply chain attack on March 31, 2026. The incident affected Ope…

Codex Security: now in research preview

OpenAI has launched Codex Security, an application security agent now available in research preview for ChatGPT Pro, Enterprise, Business, and Edu customers via Codex web. The tool addresses a critica…

Helping developers build safer AI experiences for teens

OpenAI introduced prompt-based teen safety policies designed to help developers implement age-appropriate protections for teens using the open-weight safety model, gpt-oss-safeguard. These policies tr…

Introducing the OpenAI Safety Fellowship

OpenAI has launched the OpenAI Safety Fellowship, a six-month pilot program (September 2026–February 2027) to fund external researchers focused on AI safety and alignment. The program targets independ…

Introducing the OpenAI Safety Bug Bounty program

OpenAI has introduced a Safety Bug Bounty program aimed at proactively identifying and mitigating risks associated with AI misuse and vulnerabilities. The program specifically targets critical threats…

OpenAI has launched a new Safety Bug Bounty program on March 25, 2026, to proactively identify and mitigate AI-specific abuse and safety risks across its products. Unlike traditional security-focused

OpenAI has introduced prompt-based teen safety policies for developers utilizing gpt-oss-safeguard, a framework designed to moderate age-specific risks in AI systems. The new policies aim to enhance s

OpenAI to acquire Astral

OpenAI announced its acquisition of Astral, a developer of widely used open-source Python tools such as uv, Ruff, and ty, to accelerate the growth of its Codex AI platform. The deal aims to integrate …

Introducing GPT‑5.4

OpenAI has launched GPT-5.4, a significant upgrade to its AI models, designed to enhance professional work across ChatGPT, the API, and Codex. The new model introduces GPT-5.4 Thinking and GPT-5.4 Pro…

Joint Statement from OpenAI and Microsoft

OpenAI and Microsoft reaffirmed the strength of their long-standing partnership in a joint statement on February 27, 2026, clarifying that recent announcements about OpenAI’s funding and new partnersh…

Scaling AI for everyone

OpenAI has announced a $110 billion investment round at a $730 billion pre-money valuation, securing $30 billion from SoftBank, $30 billion from NVIDIA, and $50 billion from Amazon. This funding, alon…

Our agreement with the Department of War

OpenAI has reached a landmark agreement with the U.S. Department of War (DoW) to deploy advanced AI systems in classified environments, introducing unprecedented safeguards to prevent misuse. The deal…

OpenAI to acquire Promptfoo

OpenAI will acquire Promptfoo, an AI security platform specializing in identifying and remediating vulnerabilities in AI systems during development. The acquisition aims to integrate Promptfoo’s techn…

Introducing GPT‑5.4 mini and nano

OpenAI introduced GPT-5.4 mini and nano, optimized small models for coding and subagent workloads. GPT-5.4 mini improves over GPT-5 mini across coding, reasoning, multimodal understanding, and tool us…

OpenAI raises $122 billion to accelerate the next phase of AI

OpenAI has raised $122 billion in a landmark funding round, valuing the company at $852 billion post-money, marking one of the largest capital infusions in technology history. This unprecedented inves…

Codex now offers pay-as-you-go pricing for teams

OpenAI has introduced pay-as-you-go pricing for Codex, a coding-focused AI tool, allowing teams on ChatGPT Business and Enterprise to add Codex-only seats without fixed fees. Usage is billed by token …

Powering product discovery in ChatGPT

ChatGPT has introduced a new shopping experience powered by the Agentic Commerce Protocol, enhancing product discovery with visually immersive features. The update enables users to perform side-by-sid…

Update on the OpenAI Foundation

The OpenAI Foundation has announced a commitment to invest at least $1 billion in initiatives aimed at curing diseases, enhancing economic opportunity, building AI resilience, and supporting community…

Creating with Sora Safely

OpenAI has introduced Sora 2 and the Sora app, positioning safety as a foundational element to address novel challenges posed by advanced video generation and a new social creation platform. The compa…

OpenAI has introduced a chain-of-thought monitoring system to detect misalignment risks in its internal coding agents during real-world deployments. This approach analyzes agent behavior to identify d

OpenAI has formalized its *Model Spec*, a public framework outlining how its AI models should behave across a wide range of user queries, balancing safety, fairness, and user freedom. Released in an u

OpenAI to acquire Astral

The OpenAI Foundation, recapitalized last fall, has announced its initial strategic investments totaling at least $1 billion across four key areas: life sciences and curing diseases, jobs and economic

Powering Product Discovery in ChatGPT

OpenAI has significantly enhanced ChatGPT’s shopping capabilities by introducing richer, more visually immersive product discovery experiences, powered by the expanded Agentic Commerce Protocol (ACP).…

OpenAI Japan announces Japan Teen Safety Blueprint to put teen safety first

OpenAI Japan has introduced the Japan Teen Safety Blueprint, a new initiative aimed at enhancing safety and well-being for teenagers using generative AI. The blueprint introduces stronger age verifica…

Contact our sales team

OpenAI has introduced ChatGPT Business, a new offering targeting small teams seeking enterprise-grade AI capabilities. The product emphasizes no training on data, unlimited use of its most advanced mo…

Cookbook

OpenAI has expanded its developer resources with a new cookbook featuring 239 entries, including guides on resilient prompt engineering, reinforcement fine-tuning, multi-tool orchestration, and voice …

Sora 2 Prompting Guide

The content provides a cookbook guide for crafting effective video prompts for Sora 2 generation. It focuses on techniques and best practices to enhance the quality and relevance of video outputs from…

Speeding up agentic workflows with WebSockets in the Responses API

OpenAI reduced agentic workflow latency in the Responses API by 40% by introducing WebSocket mode, replacing synchronous HTTP calls with persistent connections. This change leverages in-memory caching…

OpenAI Newsroom | AI Adoption | OpenAI

The content provided is minimal and does not offer a detailed summary. It only mentions 'AI Adoption' without any additional context or information.

OpenAI named a Leader in enterprise coding agents by Gartner

OpenAI’s Codex was recognized as a Leader in Gartner’s 2026 Magic Quadrant for Enterprise AI Coding Agents, citing its agentic capabilities, governance features, and broad developer surface. Recent up…

Using GPT-5.2

OpenAI has introduced GPT-5.2, the latest model in the GPT-5 family, which replaces GPT-5.1. GPT-5.2 is designed for both general and agentic tasks, offering improvements in accuracy and performance. …

Realtime API

The OpenAI Realtime API has transitioned from beta to general availability (GA), introducing several changes to its interface. The API now supports low-latency communication with models that handle sp…

Frontend coding with GPT-5

OpenAI's GPT-5 has demonstrated significant advancements in frontend development, showcasing its ability to create full-stack applications, perform complex refactors, and make precise edits within lar…

Text generation

OpenAI has introduced the Responses API as the recommended choice for new projects, replacing the older Chat Completions API. The Responses API offers enhanced capabilities, particularly for reasoning…

Images and vision

OpenAI has introduced gpt-image-1, a natively multimodal large language model that can process and generate images with improved instruction following and contextual awareness. This model differs from…

Image generation

OpenAI has introduced GPT Image, a new family of models for image generation and editing, which includes gpt-image-1.5, gpt-image-1, and gpt-image-1-mini. The latest model, gpt-image-1.5, is recommend…

Audio and speech

OpenAI has expanded its audio capabilities through the OpenAI API, offering various audio processing features. The API supports voice agents that can understand and respond to audio inputs, with optio…

Agents

OpenAI has introduced AgentKit, a modular toolkit for building, deploying, and optimizing agents. Agents are systems designed to accomplish tasks ranging from simple goals to complex workflows. The Ag…

OpenAI Newsroom | Recent news | OpenAI

No specific news or announcements provided.

Reasoning models

OpenAI has introduced reasoning models, including GPT-5, which are designed to perform complex problem-solving, coding, scientific reasoning, and multi-step planning. These models generate a chain of …

OpenAI Newsroom | Research | OpenAI

The page does not contain any news or announcements.

OpenAI Newsroom | Product | OpenAI

The content provided is incomplete and does not contain sufficient information to generate a summary.

OpenAI Newsroom | Safety | OpenAI

NONE

OpenAI Newsroom | Engineering | OpenAI

The webpage titled 'Engineering' on the OpenAI website focuses on stories about the technology and builders at OpenAI. It serves as a section dedicated to sharing insights and updates related to the e…

Security | OpenAI

OpenAI Newsroom | Global Affairs | OpenAI

The page is a static page with no news or announcements.

GPT‑5.1: A smarter, more conversational ChatGPT

OpenAI has released GPT-5.1, an upgrade to the GPT-5 series, which is rolling out to users starting today. The new version focuses on improving both intelligence and communication style, making it mor…

Introducing GPT‑5.2

OpenAI has introduced GPT-5.2, a significant upgrade to its AI model series, designed to enhance professional knowledge work. The new model outperforms industry professionals in various benchmarks, in…