OpenAI DevDay 2025: Seven Major Announcements You Need to Know
OpenAI just wrapped up DevDay 2025 with a wave of groundbreaking AI updates that could reshape how developers, businesses, and everyday users interact with artificial intelligence. This year’s announcements touched on every corner of the AI landscape, developer tooling, consumer experiences, advanced content generation, and the massive infrastructure that makes it all possible. From powerful new developer tools to strategic industry partnerships, the updates point toward a future where AI is more integrated, capable, and accessible than ever before.
Unlike previous years, DevDay 2025 wasn’t simply a showcase of cutting-edge models, it was an unveiling of a broader vision. OpenAI is working to create an interconnected AI ecosystem that empowers anyone, whether they’re a seasoned engineer or a first-time user, to build, deploy, and leverage AI at scale. The company made it clear that its mission is not just technological advancement, but alignment with real-world needs: efficiency, affordability, and versatility.
The momentum is undeniable. ChatGPT has now reached over 800 million weekly active users, growing by more than 50 million new weekly users every month since February 2025—a surge that reflects both the global appetite for AI and the utility of OpenAI’s offerings in daily life. On top of that, its newly announced partnership with AMD aims to deliver 6 gigawatts of computing power, laying the groundwork for future models that will dwarf today’s capabilities in speed, scale, and intelligence.
In essence, DevDay 2025 signals a shift: AI is no longer a specialized tool tucked away in labs, it’s becoming part of the world’s operational fabric. OpenAI’s updates show a commitment to driving AI adoption into every workflow, every industry, and every corner of the globe.
Here’s a breakdown of the seven biggest announcements from OpenAI DevDay 2025.
1. AgentKit: Build AI Agents with Ease
2. Apps Inside ChatGPT
3. Codex Upgrades for Developers
4. Sora 2 in the API
5. GPT-5 Pro API Access
6. New Cost-Efficient AI Models
7. Massive Compute Partnership with AMD
The Bigger Picture
1. AgentKit: Build AI Agents with Ease
What’s new: OpenAI unveiled AgentKit, a powerful new toolkit that makes it possible for anyone to build, deploy, and manage AI agents without needing deep technical expertise. These agents are designed to automate everyday workflows—handling anything from summarizing reports and scheduling meetings to analyzing data or even managing customer support conversations autonomously.
Why it matters: AgentKit represents a major leap toward democratizing AI. Until now, deploying intelligent agents required significant development time and backend orchestration. With AgentKit, OpenAI has packaged the complex logic, reasoning, and execution layers of agent design into a single developer-friendly environment. It’s not just about writing prompts—it’s about creating self-directed digital assistants that can connect with tools, APIs, and real-world systems in just minutes. Businesses can now launch specialized AI agents custom-built for their teams and operations, marking the next frontier of workflow automation.
2. Apps Inside ChatGPT
What’s new: OpenAI announced that developers can now integrate third-party apps directly inside ChatGPT, making the chat interface a dynamic hub for interacting with software and services. This means users can perform tasks like booking hotels via Airbnb, managing tasks through Notion, or shopping online—all without leaving the chat window.
Why it matters: This transforms ChatGPT from a conversational model into an AI super-app. The integration of apps introduces a new paradigm where conversation becomes the user interface. Instead of switching between multiple tabs and tools, users can simply ask ChatGPT to handle the job. For developers, this is a golden opportunity to bring their products to millions of active ChatGPT users in an intuitive, no-friction environment. It pushes the boundary between chat-based assistance and true AI-driven execution, paving the way for a world of context-aware, conversational computing.
3. Codex Upgrades for Developers
What’s new: The Codex model—the AI engine behind automated code generation—has received a major overhaul. The upgraded version introduces lightning-fast performance, improved understanding of complex programming patterns, and the ability to autonomously write, debug, and even deploy production-grade code.
Why it matters: For developers, Codex is evolving from a coding helper into a co-engineer. It can now manage end-to-end software tasks, recommend optimizations, and even detect issues before deployment. This enhancement drastically reduces repetitive work and shortens development cycles, allowing engineers to focus on higher-level design and innovation. With these upgrades, OpenAI is positioning Codex as a cornerstone of intelligent software development—bringing us closer to a future where human creativity and machine precision work seamlessly side by side.
4. Sora 2 in the API
What’s new: OpenAI has made a version of Sora 2—its highly advanced video generation model—available through the API. Sora 2 is capable of producing realistic, physics-accurate videos that simulate movement, lighting, and environmental effects with uncanny detail. Developers can now embed this technology directly into their own applications and platforms.
Why it matters: Video generation is one of the most resource-intensive AI challenges, but Sora 2’s API integration makes it accessible and scalable. Beyond entertainment, its applications in education, training simulations, product demonstrations, and scientific visualization are enormous. For example, an edtech app could generate interactive lab experiments in real time, while a marketing team could instantly produce high-quality ad videos without a filming crew. Sora 2 moves AI video from novelty to practical utility, allowing innovators to build immersive, dynamic content with minimal resources.
5. GPT-5 Pro API Access
What’s new: OpenAI is offering developers direct API access to GPT-5 Pro, a more capable and context-aware version of its flagship language model. GPT-5 Pro brings enhanced reasoning, longer context handling, and richer output customization options, making it ideal for complex, domain-specific tasks.
Why it matters: This update turns GPT-5 Pro into a high-performance backend for apps that require nuanced understanding and tailored responses—such as legal document analysis, financial forecasting, technical writing, or advanced customer engagement systems. Developers can now build software that behaves less like a static querying system and more like a dynamic expert, capable of adapting tone, style, and reasoning to fit specialized workflows. It firmly positions GPT-5 Pro as the premium standard for AI-as-a-service in enterprise and creative sectors.
6. New Cost-Efficient AI Models
What’s new:
gpt-realtime-mini: An affordable voice model producing natural, expressive speech in real time.
gpt-image-1-mini: A cost-efficient image generator that can create realistic photos and even interpret screenshots.
Why it matters: Until now, many advanced AI models were financially prohibitive for widespread use. These “mini” versions significantly lower the cost while still delivering strong performance, opening the door to AI integration for small businesses, independent developers, educators, and hobbyists. This means voice and image generation—once a luxury feature—can now be embedded into virtually any application, from mobile games to personalized learning platforms, without breaking budgets.
7. Massive Compute Partnership with AMD
What’s new: OpenAI and AMD are joining forces to deploy 6 gigawatts of AI computation capacity, starting with 1 GW of AMD Instinct MI450 GPUs in the second half of 2026. As part of the agreement, AMD has given OpenAI a warrant to acquire up to 160 million AMD shares at $0.01 each if performance goals are met—potentially making OpenAI a 10% owner of AMD.
Why it matters: This partnership represents one of the largest computing investments in AI history, ensuring that OpenAI and its developer ecosystem have the processing muscle needed for next-generation models. More compute means faster training, better model accuracy, and the ability to serve billions of high-quality AI interactions simultaneously. For developers, this ensures stability, scalability, and performance even as AI workloads skyrocket—a foundational step in enabling a truly global AI infrastructure.
The Bigger Picture
Sam Altman’s announcement of 800 million weekly ChatGPT users—with an average gain of 50 million more each month since February 2025—shows that AI adoption isn’t slowing down. OpenAI isn’t just capitalizing on this momentum; it’s building the infrastructure, ecosystem, and tools to ensure AI becomes part of everything we do.
The highlights from DevDay show a clear direction:
Lowering barriers for developers to build with AI.
Making AI more accessible and affordable.
Expanding real-world integration through apps, media generation, and voice capabilities.
Ensuring the computing backbone can handle global demand.
With these announcements, OpenAI is positioning itself not just as a technology provider, but as the engine driving AI’s mainstream integration. The real impact will come in the months ahead as these tools roll out and innovators start pushing them to their limits.