OpenAI has launched o1-pro, an upgraded version of its o1 “reasoning” AI model. This new model is now available in the developer API and it promises better performance. However, it comes with a hefty price tag. Will developers be willing to pay for it?
What Makes o1-Pro Different?
Unlike its predecessor, o1-pro uses more computing power. According to OpenAI, this extra power helps it think harder and provide more reliable answers. This model is a response for the call for models that tackle complex reasoning tasks with better accuracy.
Right now, not every developer can access it. To qualify, they must have spent at least $5 on OpenAI API services. But the real shocker? The price.
A Price Tag That Turns Heads
- $150 per million input tokens (~750,000 words)
- $600 per million output tokens
This is twice the price of OpenAI’s GPT-4.5 for input. For output, it costs ten times more than the standard o1 model. OpenAI is betting on performance over affordability.
Also read: ChatGPT Pro Is Here With a $200 Subscription
Is o1-Pro Worth the Price?
OpenAI claims that o1-pro delivers more precise responses. But early adopters have mixed opinions.
The Performance Debate
The model has been available to ChatGPT Pro subscribers since December. This gave users a preview of its capabilities. However, feedback hasn’t been all positive.
- Struggles with Sudoku: Users found that o1-pro had trouble solving Sudoku puzzles.
- Fails Optical Illusions: It got confused by simple trick questions and visual puzzles.
- Minor Coding & Math Gains: OpenAI’s own benchmarks show only slight improvements in coding and math.
Where o1-Pro Shines
Despite these flaws, OpenAI says the model is more reliable for difficult questions. This could be useful for businesses and researchers who need consistent, high-quality answers.
High-Priced AI
The release of o1-pro raises big questions about AI pricing. OpenAI seems to be targeting enterprise users rather than casual ones. But will companies see enough value to justify the cost?
What This Means for Developers
For those who need high accuracy, the price might be worth it. But for general tasks, more affordable models like GPT-4.5 could be a better choice.
