After releasing GPT-6.2, OpenAI says the cost of using specialist models has fallen to roughly one-sixth that of the previous generation. This shows that AI competition is shifting away from parameter counts and leaderboard rankings toward the question of who can complete real work reliably at lower cost. Notably, a $5 billion AI drug development financing deal and progress in agent payments emerged on the same day, indicating that industry momentum has extended beyond model capabilities into commercial deployment.

What this is

GPT-6.2 comes in two product lines: Astra targets complex projects in science, law and finance, while Sol emphasizes coding, writing and multi-step agent workflows. Agents—AI systems capable of autonomously invoking tools and completing tasks continuously—are also gaining payment capabilities. India plans to allow them to handle small automated payments, while China UnionPay has already released an open protocol covering the area.

On the capital side, Isomorphic Labs has secured $5 billion in financing, while robotics company Genesis AI has reached a valuation of $3.5 billion. Goldman Sachs has also raised its forecast for global AI project financing in 2027 to $545 billion.

Industry view

The decline in costs should expand the range of AI applications in professional services and enterprise workflows. Robin Li, chairman of Baidu, argued at the APEC summit that AI's value ultimately comes from the application layer. The industrial positioning of figures such as Kai-Fu Lee and Demis Hassabis also shows that capital is betting on measurable commercial outcomes in drug discovery, robotics and other sectors.

The risks are equally clear. California has sued OpenAI over alleged model loss of control and agent-driven cyberattacks. Research involving Yoshua Bengio has also found that understanding a model's internal mechanisms does not necessarily make it safer. Financing structures present another challenge: Goldman Sachs has warned that some AI capital expenditure is being routed through special-purpose entities—project vehicles designed to isolate risk and simplify funding—creating very limited transparency into off-balance-sheet liabilities. Rapid growth in financing does not mean cash flow and project returns have been validated.

Impact on regular people

For enterprise IT: Lower prices for specialist models make it easier to integrate AI into contract review, financial analysis and R&D workflows. At the same time, businesses need to establish corresponding access controls, audit trails and human-review mechanisms.

For individual careers: AI will continue to take on coding, writing and research organization. Repetitive copywriting and entry-level analysis roles will face greater pressure to adapt, making reviewable professional judgment more important.

For consumer markets: If agent payments develop under unified rules, AI could increasingly make choices for users across e-commerce, travel booking and bill payment, with price comparison and purchase authorization handled more frequently by machines.