OpenAI Plans $20,000 Monthly AI Agents While Chinese Startup manus.im Outperforms Them on Benchmark Tests
In the high-stakes world of artificial intelligence, being first to market isn't always enough. Sometimes, the tortoise really does beat the hare.
Last week, reports surfaced that OpenAI—the company behind ChatGPT—plans to charge enterprises up to $20,000 monthly for specialized AI "agents." But while OpenAI has been quietly developing these premium offerings, a lesser-known competitor called manus.im has emerged with what early benchmark tests suggest is superior technology at a fraction of the cost.
This development throws a significant wrench into OpenAI's ambitious pricing strategy—one the company desperately needs to succeed after reportedly hemorrhaging $5 billion last year.
OpenAI's High-Stakes Pricing Gamble
According to reporting from The Information, OpenAI intends to launch a series of specialized AI agents targeting different enterprise needs and price points:
- A "high-income knowledge worker" agent at $2,000 monthly
- A software developer agent commanding $10,000 monthly
- A premium "PhD-level research" agent priced at an eye-watering $20,000 per month
These figures represent a dramatic departure from OpenAI's current premium subscription, which tops out at $200 monthly for ChatGPT Pro. The company is betting big that enterprises will pay these substantial sums for AI assistants specialized in tasks like sorting sales leads, engineering software, and conducting advanced research.
This is a watershed moment in enterprise AI pricing. OpenAI is essentially saying their specialized agents deliver value equivalent to multiple skilled human employees—and they're pricing accordingly.
SoftBank, a major OpenAI investor, appears to believe in this vision. The Japanese conglomerate has reportedly committed a staggering $3 billion to OpenAI's agent products this year alone, signaling strong confidence in the strategy.
Enter the Disruptor: manus.im
While OpenAI has been crafting its premium pricing structure, manus.im has quietly built what might be a better alternative. The Chinese startup has developed AI agents capable of handling a remarkably similar range of tasks—from travel planning and stock analysis to educational content creation and supplier sourcing.
What's particularly troubling for OpenAI is manus.im's performance on the GAIA benchmark, widely considered the gold standard for evaluating general AI assistants. The data shows manus.im achieving state-of-the-art results across all difficulty levels:
- Level 1: 86.5% pass rate (vs. OpenAI Deep Research's 74.3%)
- Level 2: 70.1% pass rate (vs. OpenAI Deep Research's 65.8%)
- Level 3: 57.7% pass rate (vs. OpenAI Deep Research's 47.6%)
The performance gap is most pronounced at Level 3—the most complex tier designed to simulate real-world challenges. This suggests manus.im's technology might be more capable of handling the exact enterprise use cases OpenAI hopes to monetize with its premium pricing.
The $5 Billion Question
OpenAI's aggressive pricing strategy comes from necessity as much as ambition. The company reportedly lost approximately $5 billion in 2023, primarily due to the massive computational costs of running its services and extensive R&D expenses.
OpenAI is stuck between a rock and a hard place. They've built incredible technology, but the economics of running it at scale are brutal. These high-priced enterprise agents represent their best shot at a sustainable business model.
The company reportedly aims for these AI agents to eventually contribute 20-25% of its total revenue—a crucial pillar in their path to profitability.
Real-World Capabilities: The Practical Test
Beyond benchmark scores, manus.im appears to excel in practical applications that mirror OpenAI's targeted use cases. The company showcases abilities like:
- Creating comprehensive travel handbooks for international trips
- Performing deep analytical dives on stock performance with visual dashboards
- Developing educational content including video presentations
- Analyzing business operations using uploaded store data
- Conducting competitive research across multiple web sources
These capabilities align remarkably well with the enterprise functions OpenAI hopes to monetize through its tiered agent pricing.
What's particularly interesting about manus.im's approach is how they've optimized for practical utility rather than just raw capability. Their agents seem designed with real-world applications as the primary focus.
Market Implications and Investor Outlook
OpenAI's pricing strategy faces significant challenges beyond just manus.im. Google and Anthropic are also developing competitive offerings, and enterprises remain cautious about committing to substantial monthly expenditures for relatively unproven technology.
The $20,000 monthly price point makes sense only if these agents can truly replace specialized knowledge workers. But that's a big 'if.' Companies will demand proof of ROI before making that level of commitment.
For investors watching this space, the emergence of manus.im introduces both risk and opportunity. OpenAI's valuation—and by extension, Microsoft's AI-focused growth narrative—partially hinges on successfully monetizing these advanced capabilities at premium price points.
Meanwhile, the benchmark superiority demonstrated by manus.im suggests we may be witnessing the emergence of a significant new player in enterprise AI—one that could potentially disrupt established players' pricing power before it's even fully established.
The Road Ahead
OpenAI hasn't announced official launch dates for its enterprise agents, and eligibility criteria remain unclear. This gives the company time to potentially refine its offerings and pricing strategy in response to competitive pressures.
What's certain is that the AI agent landscape is evolving rapidly. For enterprises considering investments in this technology, the emergence of manus.im as a credible competitor to OpenAI could ultimately lead to more options and potentially more favorable pricing.
As the dust settles, one thing becomes increasingly clear: OpenAI's path to profitability just got considerably more complicated. The company that popularized modern generative AI now finds itself in an unfamiliar position—playing catch-up to a nimble competitor that may have beaten them at their own game.