OpenAI's GPT-5 Underwhelms Consumers, But Scores Big with Enterprises

GPT-5 receives average reviews from ordinary users

CNBC reported that OpenAI's latest flagship model, GPT-5, has finally been released, but it has received lukewarm response from the general public. However, for OpenAI CEO Sam Altman, the good news is that his big model is doing well in the lucrative enterprise market.

Last week, GPT-5's release was met with some criticism. Critics complained about its lack of intuitive interaction, ultimately forcing OpenAI to reopen the old version GPT-4 model for paying users.

However, GPT-5's focus is not on consumer markets but on capturing the enterprise market. In this area, OpenAI's competitor Anthropic has already taken the lead.

Enterprise praise

Just one week after its release, Cursor, Vercel, and Factory, among other startup companies, have announced that they will set GPT-5 as their default model for certain key products and tools. They praised it for being faster to deploy, having better performance on complex tasks, and being more affordable.

Several companies said that in the area of code design and interface where Anthropic had taken the lead, GPT-5 has now caught up or even surpassed Claude, the top model from Anthropic.

Box is testing GPT-5 for its CEO Aaron Levie, who told CNBC that this model is a "breakthrough" with reasoning abilities that surpass those of previous systems.

Behind the scenes, OpenAI has established its own enterprise sales team, led by Chief Operating Officer Brad Lightcap, which has over 500 people and operates independently of Microsoft. Microsoft has been OpenAI's main investor and key cloud service partner. Customers can access GPT models through Microsoft Azure or directly through OpenAI. OpenAI controls the API and product experience.

However, the economic cost is extremely high. The large model running costs are enormous, with OpenAI and Anthropic both investing heavily to lock in customers, with OpenAI expected to burn around $8 billion this year.

New clients

GPT-5's price is significantly lower than Anthropic's top model Claude Opus 4.1, which is sometimes even 7.5 times cheaper in certain cases. However, OpenAI is investing heavily to build the infrastructure needed to maintain this price advantage.

For OpenAI, the focus now is on acquiring clients, locking in users, and building a real business model based on user loyalty.

Cursor is currently still one of Anthropic's big clients but is now directing new users to OpenAI. Cursor co-founder and CEO Michael Truell emphasized this change during OpenAI's live launch event, calling GPT-5 "the most intelligent programming model we've tried."

OpenAI is competing with Anthropic in the enterprise market

However, Truell said that this change only applies to new registered users, while existing Cursor clients will continue using Anthropic's products as their default model. Cursor and Anthropic have signed long-term contracts with income guarantees, with the latter building its business model on locking in enterprise-level markets.

An anonymous insider revealed that as of June, enterprise revenue accounted for approximately 80% of Anthropic's income, with year-over-year growth of 16 times. The source said that the company's revenue has increased by $3 billion over the past six months, reaching a total of $10 billion in June alone. This year, the number of eight- and nine-figure contracts signed is already three times higher than the whole-year total for 2024.

Anthropic said that its enterprise business covers areas far beyond the tech sector.

Although Anthropic has a wide range, OpenAI is rapidly expanding in the enterprise market. An insider claimed that GPT-5's API call volume has surged since its release, with current processing and smart body construction tasks increasing by over two times, while complex reasoning cases have increased by more than 7 times. Enterprise demand is rising rapidly, especially for planning and multi-step reasoning tasks.

High value

The developer platform Factory said that they are closely working with OpenAI to set GPT-5 as their default model.

Factory CEO Matan Grinberg said, "In setting up plans and executing complex programming solutions, GPT-5 is much better. Its long-term planning ability and scheme continuity are both more impressive." He pointed out that GPT-5 has excellent compatibility with their multi-intelligent body platform, saying it can perfectly coordinate the high-level design details with low-level implementation details.

Factory set GPT-5 as its default model for another important reason: price flexibility.

"Price is what our end-users care about most." Grinberg said. He added that the lower reasoning cost allows customers to be more willing to try new ideas. Users no longer need to repeatedly consider whether a question is worth spending money on; they can "try things out more freely," without hesitation, exploring new ideas.

Lovable co-founder and CEO Anton Osika said that before GPT-5 was officially released, their team had spent several weeks testing it and were extremely satisfied with its performance. The company develops AI tools that allow users to create real software businesses without writing code.

"We found the new model is much stronger in many complex application scenarios." Osika emphasized. He stressed that GPT-5 "is more inclined to take action and reflect on its behavior," and will invest more time ensuring the accuracy of decisions. (Author/Ai Yu)