OpenAI is reportedly preparing a new class of AI agents with prices far above the consumer subscription levels most ChatGPT users know today. The planned products, described in a report by The Information and based on information shared with investors, would target professional users and organizations willing to pay for specialized automation.
The reported pricing is striking: $2,000 per month for basic agents, $10,000 per month for developer agents, and $20,000 per month for Ph.D.-level research assistants. If launched as described, these OpenAI AI agents would mark a sharp move toward high-end AI services built around work replacement, not only general chatbot access.
Three reported tiers for professional AI agents
The Information reports that OpenAI is planning three separate agent products. Each tier appears aimed at a different kind of work, with pricing that reflects the value OpenAI says these systems could provide.
- Basic agents: reportedly priced at $2,000 per month and aimed at "high-income knowledge workers".
- Developer agents: reportedly priced at $10,000 per month and designed as coding assistants for experienced programmers.
- Ph.D.-level research assistants: reportedly priced at $20,000 per month and intended for complex scientific research questions.
The examples given are practical rather than abstract. Basic agents would handle work such as evaluating sales leads. Developer agents would support experienced programmers. The highest tier would focus on difficult scientific research questions.
That structure shows how OpenAI may be thinking about the AI services market. Instead of selling one broad subscription for many users, the company could segment AI agents by job category, expected productivity, and the cost of the human work they are meant to replace.
A bet on replacing expensive specialist work
The report says OpenAI is positioning the agents as direct replacements for highly-paid specialists. That framing matters because it explains why the monthly fees are so high. A $20,000 per month AI assistant would not be marketed like a normal software subscription; it would be sold against the cost of specialized professional labor.
For companies, the buying decision would likely depend on whether these agents can handle valuable tasks reliably enough to justify the price. The source article gives three examples of those tasks: sales lead evaluation, software coding support, and complex scientific research questions.
Those examples also show why the agent label is important. The products are not described only as tools that answer prompts. They are framed around doing defined work for professional users, which is why OpenAI could attach much higher prices than a standard ChatGPT plan.
Why the revenue stakes are high
OpenAI reportedly expects these agent products to generate between 20 and 25 percent of its total revenue. That would make the agent line more than an experiment if the plans move forward as described.
The company has also secured financial backing tied to this direction. Softbank has committed $3 billion toward OpenAI agents this year, according to the source article.
That combination of premium pricing and investor-facing revenue expectations suggests OpenAI sees AI agents as a major commercial category. The pricing tiers also point to a market where the highest-value use cases may come from professional workflows rather than broad consumer usage.
The pricing push comes with profitability pressure
The reported agent plans arrive alongside a difficult financial backdrop. The source article says OpenAI does not expect to reach profitability until 2030. It also says ChatGPT Pro, currently priced at $200 per month, operates at a loss, according to CEO Sam Altman.
That detail is important because it shows why OpenAI may be looking beyond flat-rate subscriptions. Heavy usage can make a fixed monthly fee costly for the company if the service is expensive to provide. Altman reportedly anticipates that 20 to 30 percent of Pro users would need to pay higher rates in the future because of heavy usage.
In that context, high-priced agents and usage-based pricing are connected ideas. Both are ways to align price more closely with the value and cost of advanced AI features.
Usage-based ChatGPT credits are also being explored
OpenAI is also exploring a usage-based pricing model, according to the source article. In early March, Altman proposed turning the current $20 ChatGPT subscription into credits that users could spend on specific features such as Deep Research, o1, or GPT-4.5.
Early user feedback has been cautious. The source article says many users expressed concern about "credit anxiety", meaning the worry that they would limit their own usage to avoid spending credits.
That concern highlights a tradeoff for OpenAI. Usage-based pricing could help match cost to consumption, but it may also change how people use ChatGPT. If users feel they must ration access to advanced features, the product experience could become less fluid than a simple subscription.
The reported AI agent pricing shows where OpenAI may be heading: higher-value products for professional work, higher fees for heavy or specialized use, and more experimentation with how advanced AI access is priced. The key question is whether customers will see enough practical value in agents to support monthly fees as high as $20,000.