GPT-5 Wins Enterprise Market Despite Consumer Criticism

OpenAI's latest AI model, GPT-5, has not been well-received by consumers, but it has garnered significant attention from the enterprise market. The model, which was released last week, faced criticism for its lack of intuitive user experience, leading the company to revert to the previous GPT-4 model for paying chatbot users. However, GPT-5 is not targeted at consumers; instead, it is part of OpenAI's efforts to penetrate the enterprise market, where competitor Anthropic has already established a strong presence.

Within a week of its release, several startups, including Cursor, Vercel, and Factory, announced that they had set GPT-5 as the default model for some of their key products and tools. These companies praised GPT-5 for its faster setup, superior performance in complex tasks, and lower cost. Some enterprises reported that GPT-5 now matches or even surpasses Anthropic's Claude in areas such as coding and interface design, where Anthropic had previously dominated.

Another enterprise client, Box, has been testing GPT-5 on lengthy, logic-intensive documents. The company's chief executive described the model as a "breakthrough," citing its unparalleled reasoning capabilities compared to previous systems. OpenAI has assembled its own enterprise sales team, led by the chief operating officer, consisting of over 500 people, operating independently of Microsoft. Microsoft has been a major investor and key cloud partner for OpenAI. Customers can access GPT models through Microsoft Azure or directly through OpenAI, which controls the API and product experience.

However, the economic benefits are not clear-cut. The operation of these models is costly, and both OpenAI and Anthropic are heavily investing to secure clients. OpenAI's expenditures for this year are projected to reach 80 billion. This is one of the reasons why both Anthropic and OpenAI are seeking new capital. OpenAI is exploring a secondary stock offering, which could value the company at around 500 billion, and has stated that the weekly active users of ChatGPT have approached 700 million. Anthropic is also seeking new funding, with a potential valuation of 170 billion.

GPT-5's pricing is significantly lower than Anthropic's high-end Claude Opus 4.1, in some cases by a factor of seven and a half. However, OpenAI is investing heavily in infrastructure to maintain this advantage. For OpenAI, this is about immediately winning customers, locking them in, and building a genuine business on that foundation. Cursor, while still a major client of Anthropic, is now guiding new users towards OpenAI. The company's co-founder and chief executive emphasized this shift during OpenAI's launch broadcast, calling GPT-5 "the most intelligent coding model we've tried."

This change applies only to new registrants, as existing Cursor clients will continue to use Anthropic as the default model. Cursor has a revenue commitment contract with Anthropic, which bases its business on dominating the enterprise layer. An insider revealed that as of June, enterprise revenue accounted for about 80% of total revenue, with year-over-year revenue growth at 17 times. The company added 30 billion in revenue over the past six months, including 10 billion in June alone, and the number of eight and nine-figure deals signed this year is three times that of the entire year 2024.

Anthropic's enterprise business extends beyond the tech sector. Claude supports tools for Amazon Prime, Alexa, and AIG, and is adopted by leading companies in pharmaceuticals, retail, aviation, and professional services. The company is integrated into Amazon Web Services, Google Cloud Platform, Snowflake, Databricks, and Palantir, with transaction volumes growing rapidly. An insider noted that over the past year, average customer spending has increased by more than five times, with over half of enterprise clients now using multiple Claude products. Excluding the two major clients, the rest of the business saw revenue grow by more than eleven times year-over-year.

Despite Anthropic's broad coverage, OpenAI is making progress with enterprise clients. An insider revealed that since its release, the usage of the GPT-5 API has surged, with the model currently handling more than double the volume of coding and agent construction work, and reasoning use cases have increased by more than eight times. Enterprise demand is growing rapidly, particularly in planning and multi-step reasoning tasks.

The popularity of GPT-5 over the past week demonstrates how quickly user loyalty can shift when performance and pricing favor OpenAI. An AI-driven coding platform, Qodo, recently compared GPT-5 with top models like Gemini 2.5, Claude Sonnet 4, and Grok 4, and in a blog post, stated that GPT-5 leads in identifying coding errors. The company noted that the model is often the only one capable of detecting critical issues, such as security vulnerabilities or damaged code, and suggesting clean, targeted repairs while skipping unchanged code. Drawbacks include occasional false positives and some redundancy.

The web application cloud platform Vercel has set GPT-5 as the default model for its new open-source "vibe coding" platform, which can convert simple English prompts into real-time usable applications. Vercel has also integrated GPT-5 into its dashboard's built-in Agent system, stating that GPT-5 excels at handling complex tasks and thinking through long instructions. The chief technology officer of Vercel noted that while there are many competitors in the AI model space, Claude has been the leader. It is the best coding model to date, far surpassing other competitors. OpenAI had no chance. GPT-5 has changed that. They have at least caught up and in some areas, they are even better, though in others, they are worse.

GPT-5 excels in early prototype and product design, and is more creative than Claude's Sonnet. Traditionally, new models require optimization, but with GPT-5, the integration has been seamless from the start. JetBrains has set GPT-5 as the default language for its AI Assistant and Kineto, a new no-code tool for building websites and applications, after finding that GPT-5 can generate simple, single-purpose tools faster based on user prompts. The developer platform Factory has also closely collaborated with OpenAI to set GPT-5 as the default language for its tools.

Factory's chief executive highlighted GPT-5's superior performance in planning and executing complex coding solutions, as well as maintaining consistency over the long term. GPT-5 integrates perfectly with their multi-agent platform, handling both high-level details and low-level implementation details flawlessly. Pricing flexibility was also a major factor in Factory's decision to default to GPT-5. The chief executive noted that pricing is the top concern for their end users, and cheaper reasoning methods make clients more willing to experiment. They can explore various ideas more freely without hesitation.

The co-founder and chief executive of Lovable, an AI-driven tool company aiming to enable anyone to create real software businesses without writing any code, stated that their team had beta-tested GPT-5 before its official release and was very satisfied with the improvements. They found it more powerful and smarter in many complex use cases. The new model is more action-oriented and reflective of its actions, investing more time to ensure accuracy. The chief executive of Box highlighted that the greatest benefit of GPT-5 is in enterprise workflows unrelated to coding. Their team tested the model for several weeks on complex real business data, from lengthy lease agreements to product roadmaps, and found that it excelled at solving problems that early AI systems struggled with.

The chief executive added that for enterprise applications, the behind-the-scenes operation of AI agents executing tasks is crucial, making GPT-5 a true breakthrough in work automation. The chief executive described GPT-5's performance as incredible, undoubtedly OpenAI's best model, and the best in many of their tests.