GPT-5's rollout fell mostly flat, but the model is gaining where it matters most

Sam Altman has transformed OpenAI into a staple in the tech landscape with ChatGPT. Now, three years later, his focus has shifted toward the lucrative enterprise sector. The recent launch of GPT-5, OpenAI's latest artificial intelligence model, encountered some challenges, eliciting criticism for its user experience. This led the company to revert to its predecessor, GPT-4, for some paid chatbot customers. However, GPT-5 is not aimed at the general consumer market; it represents OpenAI's strategic move to penetrate the enterprise domain, an area where competitors like Anthropic have already established a foothold. In just one week post-launch, companies such as Cursor, Vercel, and Factory have adopted GPT-5 as their preferred model for critical applications, highlighting its quick setup, enhanced performance on intricate tasks, and competitive pricing. Some businesses claim that GPT-5 now rivals or surpasses Anthropic's Claude in areas like coding and interface design—a sector that Anthropic previously led. Box, another enterprise user, has been experimenting with GPT-5 on lengthy, logic-intensive documents. CEO Aaron Levie described the model as a "breakthrough," noting its superior reasoning abilities compared to earlier AI systems. In the background, OpenAI has expanded its enterprise sales team to over 500 personnel, led by COO Brad Lightcap. This team operates independently of Microsoft, which has been a significant investor and cloud service partner for OpenAI. Customers can opt to access GPT models either through Microsoft Azure or directly from OpenAI, which manages the API and user experience. Despite the promising advancements, the economics of running these models are challenging. Both OpenAI and Anthropic are investing heavily to secure customers, with OpenAI projected to incur $8 billion in losses this year alone. This financial strain has prompted both companies to seek additional funding, with OpenAI considering a secondary stock sale that could value it around $500 billion. As of now, GPT-5 is significantly more affordable than Anthropic's top-tier Claude Opus 4.1—by as much as seven and a half times in some cases—yet OpenAI is heavily investing in infrastructure to maintain this competitive advantage. The goal is to attract customers quickly, foster loyalty, and build a sustainable business model. Cursor, which remains a major client of Anthropic, is now directing its new users toward OpenAI. Co-founder and CEO Michael Truell emphasized this shift during OpenAI's launch event, calling GPT-5 the "smartest coding model we've ever tried." This transition applies only to new users, as existing Cursor customers will continue using Anthropic's models due to contractual obligations. As of June, enterprise revenues comprised approximately 80% of Anthropic's total income, with an annualized growth rate of 17 times year-over-year, according to a source who spoke on the condition of anonymity. The company has seen a remarkable $3 billion in revenue growth over just the past six months, including $1 billion in June alone. Even with Anthropic's extensive reach across various sectors, OpenAI is making significant gains among enterprise customers. Following the launch of GPT-5, API usage has soared, with the model handling over double the amount of coding and agent-building tasks and an eightfold increase in reasoning applications. Several companies have reported that GPT-5 outperforms its competitors in identifying coding mistakes and suggesting effective, focused solutions. Vercel has integrated GPT-5 into its new open-source platform, facilitating the transformation of plain-English prompts into functional applications, and has found its performance particularly adept at managing complex tasks. JetBrains has also adopted GPT-5 as the default in its AI Assistant and Kineto—a no-code tool for web and app development—after observing its ability to quickly generate useful tools based on user prompts. Factory has worked closely with OpenAI to integrate GPT-5 into its own tools, highlighting improvements in planning and coherence over extended projects, which are crucial for their multi-agent platform. Moreover, pricing flexibility was a significant factor in Factory's decision to adopt GPT-5 as its standard model. With lower inference costs, users can experiment more freely, allowing for greater innovation without the fear of high costs. Feedback from companies like Lovable, which creates AI tools for software development, has been overwhelmingly positive. They noted that GPT-5 demonstrates enhanced capabilities in complex scenarios, making it more effective in executing tasks and reflecting on its actions. Levie from Box reiterated that the most substantial improvements are evident in enterprise workflows unrelated to coding, showcasing GPT-5's ability to tackle complex business documents and real-world data effectively. This performance could potentially make GPT-5 a game-changer for workplace automation, as it has demonstrated exceptional capabilities that set it apart from previous models.