OpenAI DevDay 2025: GPT-5 Pro, Sora 2 & Platform Updates

OpenAI DevDay 2025: Major Announcements (October 6, 2025)
On October 6, 2025, OpenAI held its third annual DevDay in San Francisco (Fort Mason), where CEO Sam Altman and team unveiled a series of new products, partnerships, and platform upgrades aimed at expanding AI’s role in software and enterprise. The announcements spanned everything from flagship models to developer tools and business collaborations. In summary, OpenAI announced:
- New AI Models: GPT-5 Pro (an advanced API model) and new real-time voice models; the next-generation video/audio generator Sora 2 ( www.digit.in) ( www.digit.in).
- Developer Tools: A suite of SDKs and toolkits – notably AgentKit (for quickly building AI agents), ChatKit (for embedding ChatGPT chat in apps), and an Apps SDK (letting third-party apps like Spotify or Zillow integrate directly with ChatGPT) ( www.digit.in).
- ChatGPT Platform Features: Integrated ChatGPT Apps and custom GPTs, so users can call named apps (e.g. generate a Spotify playlist or filter Zillow listings) directly inside ChatGPT ( www.axios.com) ( m.economictimes.com). ChatGPT will also proactively suggest relevant apps during conversations.
- Partnerships & Infrastructure: Collaborations with companies like Spotify, Zillow, and Mattel to integrate OpenAI’s tools in their services ( www.reuters.com) ( www.reuters.com), and a multibillion-dollar chip partnership with AMD to supply high-performance AI GPUs (with OpenAI having an option to purchase up to a 10% stake in AMD) ( www.reuters.com).
Each of these announcements reflects OpenAI’s strategy to turn ChatGPT into a general-purpose AI platform and to push AI into all software workflows.
ChatGPT: From Chatbot to AI Platform
OpenAI emphasized that ** ChatGPT is evolving beyond a chatbot into a full-fledged AI platform or “operating system.”** ChatGPT now serves 800 million weekly active users (up from 100M in 2023) and supports 4 million developers using its APIs ( m.economictimes.com) ( m.economictimes.com). To leverage this scale, OpenAI announced a new Apps SDK (in preview) so developers can build apps that run directly inside ChatGPT ( m.economictimes.com) ( m.economictimes.com). For example, users can now invoke apps like Spotify and Zillow from within a chat – generating music playlists or filtering real estate listings – without leaving the ChatGPT interface ( www.axios.com) ( m.economictimes.com).
As Axios reported, OpenAI’s goal is to make it easy for any app to connect to users through ChatGPT. On DevDay the company “rebooted” its ChatGPT app effort, introducing a streamlined strategy for custom GPTs and app integrations ( www.axios.com). Custom GPTs (like specialized apps) can now be activated simply by name during a chat, and ChatGPT will proactively suggest relevant apps for the task at hand ( www.axios.com). To support this, OpenAI unveiled updated developer guidelines and tools for app creators, with plans to roll out more apps later in the year. These changes “mark a significant evolution” in the ChatGPT ecosystem, strengthening the link between developers and users ( www.axios.com).
This shift is part of a broader vision to embed ChatGPT everywhere. OpenAI executives likened ChatGPT to an AI “operating system” guiding user-device interaction ( www.axios.com). As ChatGPT head Nick Turley noted, the company imagines the chatbot as a central interface – akin to how a smartphone OS works – eventually powering smart glasses or new hardware as well ( www.axios.com). Although OpenAI isn’t killing traditional interfaces, it’s betting that integrating apps and tools into ChatGPT will make it an indispensable productivity and commerce hub ( www.axios.com) ( m.economictimes.com).
New Models and Capabilities
OpenAI unveiled several new models and model updates at DevDay. Chief among them was ** GPT-5 Pro**, described as a more powerful version of GPT-5 optimized for developers ( www.digit.in). GPT-5 Pro offers “extended reasoning” and uses additional compute on complex problems, making it suited for advanced enterprise and research use cases ( www.digit.in). Importantly, OpenAI said GPT-5 Pro is now available via the API ( www.digit.in), giving developers access to its heightened capabilities. This is the first time OpenAI has offered developers access to its very top-end reasoning model in production.
In parallel, OpenAI introduced new real-time voice models. A new miniaturized voice AI, dubbed gpt-realtime-mini, was announced ( www.digit.in). This model enables interactive voice conversations: it can listen and respond in real time, facilitating applications like automated assistants or chatbots that talk. (Indian Express reports noted two new “cheaper” real-time voice models were unveiled at DevDay, though GPT-realtime-mini was specifically highlighted by OpenAI, targeting on-device or low-latency use) ( www.digit.in). These voice models complement OpenAI’s prior GPT-4o model by offering cost-effective, streamlined voice capabilities.
Another headline was Sora 2, OpenAI’s latest video-and-audio generation model ( www.digit.in). Sora 2 “is here” after being released just days earlier (Sep 30), and at DevDay OpenAI noted its availability via API. The upgraded Sora 2 model adds realistic soundscapes and synchronized dialogue to video – for example, characters in a generated scene can now have synchronized speech and natural sound effects ( www.digit.in). In essence, Sora 2 represents a leap for text-to-video akin to GPT-3.5 for language. Mattel’s partnership (below) showcases one use case: transforming toy design sketches into animated videos. By offering Sora 2 through its platform, OpenAI is enabling developers to create rich multimedia content (storyboards, animations, etc.) with ease ( www.digit.in).
In addition, OpenAI mentioned rolling out cost-effective smaller models (like versions of GPT-4o for voice) and hinted at model distillation and caching to lower latency and costs. (For example, realtime API upgrades were listed as a highlight in an AI news summary.) But the key takeaway was the breadth of model capabilities on offer: text (GPT-5), voice, vision, agents – signaling OpenAI’s aim to cover all modalities.
New Developer Tools and SDKs
To help developers build AI-powered applications faster, OpenAI announced several new developer toolkits. The centerpiece is AgentKit, described as a toolkit that lets developers “design and deploy AI agents in minutes” ( www.digit.in). Agents are autonomous programs that can perform tasks given instructions – e.g. booking travel, analyzing data, etc. With AgentKit, developers can rapidly configure agents that use GPT-5 Pro under the hood. For example, a travel agent that uses knowledge of user’s calendar and preferences, or a sales agent that automatically drafts proposals. AgentKit was expressly highlighted as a way to simplify the creation of AI-driven workflows ( www.digit.in).
OpenAI also introduced ChatKit, a toolkit for integrating ChatGPT’s chat interface into third-party apps ( www.digit.in). With ChatKit, an app developer can embed a ChatGPT-powered chat box or assistant directly inside their app or website, complete with custom prompts or workflows. For instance, a language-learning app could embed ChatGPT for conversational practice, or a productivity app could offer a ChatGPT helper in the corner.
Moreover, an Apps SDK (also called ChatGPT Apps SDK) was rolled out. This SDK allows existing apps like Canva or Spotify to plug into the ChatGPT ecosystem ( www.digit.in). In practice, it means developers of popular services can expose specific functionality (e.g. a Canva graphics generator or a Zillow listings filter) as “apps” that ChatGPT can call. During the demo, OpenAI showed logos of apps like Spotify and Zillow making ChatGPT more than just text; for example, OpenAI’s CEO Sam Altman mentioned generating a music playlist by voice command in ChatGPT. The SDK is in preview for developers, enabling them to build apps that connect to data, trigger actions, and include interactive interfaces ( m.economictimes.com).
These toolkits round out OpenAI’s platform approach: not only does OpenAI supply raw models and APIs, but it also provides higher-level frameworks (AgentKit, chat and app SDKs) so that developers can integrate AI into products with minimal effort ( www.digit.in) ( m.economictimes.com).
Strategic Partnerships and Enterprise Focus
OpenAI used DevDay to announce a number of high-profile partnerships across industry, reflecting a strategic shift toward enterprise integration ( www.reuters.com). Notably, OpenAI partnered with Spotify (for music), Zillow (real estate), and others to “integrate OpenAI’s AI tools across multiple industries” ( www.reuters.com). These partnerships mean that the services of those companies will deeply incorporate OpenAI models – for instance, Zillow can use ChatGPT to automatically organize and filter property listings based on chat queries, or Spotify can let users have conversational playlist creation with ChatGPT.
Consumer brands were also featured: toy-maker Mattel announced it will use OpenAI’s Sora 2 model to accelerate product development ( www.reuters.com). In this collaboration, Mattel designers can upload toy sketches and have Sora 2 generate animated, detailed video concepts. This lets stakeholders see a rich preview of a toy design before physical prototyping; as Reuters noted, it “demonstrates how AI can accelerate product development in the toy industry” ( www.reuters.com). Other companies (like design tool Figma, listed in news coverage) also saw their stock prices rise on news of partnerships, indicating broad enthusiasm around OpenAI’s enterprise push ( www.reuters.com).
The most significant infrastructure deal announced was with AMD. OpenAI entered a multi-year chip-supply partnership with AMD to support the company’s massive compute needs ( www.reuters.com). Under this agreement, AMD will provide up to 6 gigawatts worth of its high-performance Instinct MI450 AI GPUs starting in late 2026 ( www.reuters.com). The deal is reportedly worth tens of billions of dollars in chips, and it even gives OpenAI an option to buy up to 10% of AMD’s stock at a fixed price ( www.reuters.com). This AMD partnership helps OpenAI diversify beyond NVIDIA (with whom it already has previous deals) and addresses the “critical bottleneck” of compute capacity mentioned by OpenAI’s leaders ( www.digit.in) ( www.reuters.com). AMD’s shares jumped over 23% on the news of the deal ( www.reuters.com).
OpenAI President Greg Brockman also said that despite big losses in prior quarters, the company is doubling down on long-term initiatives like its $1 trillion “Stargate” data center buildout. The focus at DevDay was clear: strengthen the “enterprise chassis.” Altman remarked that parts of the industry are “bubbly,” but affirmed that genuine value is being created and that OpenAI is building a top-tier enterprise platform ( www.reuters.com).
Growth Metrics and Impact
Altman glazed the audience with OpenAI’s growth metrics. He announced ChatGPT now has 800 million weekly users (up from 100M just two years ago) ( m.economictimes.com). The developer ecosystem has also exploded – the number of developers using OpenAI’s tools has doubled to 4 million ( m.economictimes.com). API usage has similarly skyrocketed: backend usage jumped from 300 million tokens per minute to over 6 billion per minute ( m.economictimes.com). These figures underscore how ubiquitous ChatGPT and its APIs have become in just a few years.
To put it in perspective, Altman noted that OpenAI has more users now than many national platforms, and its roadmap is heavily oriented around scaling those users with new features. (For example, the ChatGPT apps and suggestions are designed to increase how much end users do inside ChatGPT.) The media buzz around DevDay emphasized that OpenAI’s platform approach is turning it into a core infrastructure player for the AI age: “ChatGPT Active User Base: 800 million weekly” was a headline on multiple tech sites ( www.digit.in).
Human-Centric Hardware (Jony Ive)
An unexpected highlight was a guest appearance by former Apple design chief Jony Ive. Ive joined Sam Altman on stage to discuss “AI-driven hardware” being developed with a focus on emotional well-being ( www.digit.in). While details were sparse, Ive said the goal is to create technology that “helps people feel happier, more peaceful, and less disconnected.” This suggests OpenAI is exploring dedicated AI devices (beyond software) designed around good design and user experience. It aligns with comments that ChatGPT might one day power new kinds of smart-glasses or other form factors. In short, OpenAI hinted that the future may include AI-native hardware, co-designed with a human-centric ethos ( www.digit.in).
Conclusion
Overall, the announcements on October 6, 2025 signal that OpenAI is aggressively expanding across multiple fronts. The company is injecting AI into more software and services (through ChatGPT Apps and partnerships), advancing core model capabilities (GPT-5 Pro, Sora 2, voice), and empowering developers (AgentKit, SDKs). It’s also cementing enterprise and infrastructure support with partnerships from media to hardware (AMD’s gigawatt deal alone is in the tens of billions ( www.reuters.com)). These moves show OpenAI is aiming to be both the foundation and the toolkit for a new generation of AI applications.
In short, DevDay 2025 was “bigger than ever,” as OpenAI itself teased beforehand ( openai.com): it delivered on that promise by unveiling a broad vision of AI as an integral part of daily software, business processes, and even future devices. The October 6 announcements make clear that OpenAI intends to shape the next wave of AI innovation for developers and consumers alike ( aibreaking.org) ( www.reuters.com).
Sources: Official content and news reports from OpenAI, Reuters, Axios, the Economic Times, and other tech media ( openai.com) ( www.reuters.com) ( www.axios.com) ( www.axios.com) ( www.reuters.com) ( www.digit.in) ( www.digit.in) ( m.economictimes.com) ( m.economictimes.com) ( www.reuters.com).
DISCLAIMER
The information contained in this document is provided for educational and informational purposes only. We make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability of the information contained herein. Any reliance you place on such information is strictly at your own risk. In no event will IntuitionLabs.ai or its representatives be liable for any loss or damage including without limitation, indirect or consequential loss or damage, or any loss or damage whatsoever arising from the use of information presented in this document. This document may contain content generated with the assistance of artificial intelligence technologies. AI-generated content may contain errors, omissions, or inaccuracies. Readers are advised to independently verify any critical information before acting upon it. All product names, logos, brands, trademarks, and registered trademarks mentioned in this document are the property of their respective owners. All company, product, and service names used in this document are for identification purposes only. Use of these names, logos, trademarks, and brands does not imply endorsement by the respective trademark holders. IntuitionLabs.ai is an AI software development company specializing in helping life-science companies implement and leverage artificial intelligence solutions. Founded in 2023 by Adrien Laurent and based in San Jose, California. This document does not constitute professional or legal advice. For specific guidance related to your business needs, please consult with appropriate qualified professionals.