Ad
Skip to content
Subscribe Now
AI in practice
Copy the url to clipboard Share this article Go to comment section

GPT-4.5 is no longer the most expensive AI model

Maximilian Schreiner
Maximilian Schreiner View the LinkedIn Profile of Maximilian Schreiner
Mar 20, 2025
Image description
OpenAI

Key Points

  • OpenAI has now made its o1 Pro model, previously available only through ChatGPT, available through the API - at significantly higher prices than the previous most expensive model, GPT-4.5: $150 per million input tokens and $600 per million output tokens.
  • O1 Pro outperformed previous models in mathematics, programming and scientific questions, but is primarily aimed at researchers, engineers and professional users with specific requirements.
  • Despite new competition such as Claude 3.7 "Thinking" or Grok 3 "Think", o1 Pro remains one of the three most powerful models and could be the best choice for certain applications due to its use of additional processing power for more precise results.

Following its initial release in late 2024, OpenAI has expanded access to o1 Pro beyond ChatGPT through its API interface.

The company's premium ChatGPT offering launched with a $200 monthly subscription - a significant jump from the standard $20 plan. According to OpenAI CEO Sam Altman, o1 Pro represents the most intelligent model in the world designed to tackle the hardest problems.

The model shows improved success rates compared to previous versions across mathematics competitions, programming tests, and scientific inquiries. While OpenAI has internally moved on to o3, that successor model remains unavailable except for its mini variant.

The API pricing for o1 Pro makes GPT-4.5 look economical by comparison: Input costs run $150 per million tokens, with output at $600. GPT-4.5 charges $75 and $150 respectively, while the standard o1 comes in at $15 and $60.

Ad
DEC_D_Incontent-1

Limited practical applications for most users

OpenAI positions the service primarily for researchers, engineers, and professional users.  Since o1's debut, several competing reasoning models have emerged, including Claude 3.7 "Thinking" and Grok 3 "Think."

However, o1 maintains its position as one of the top three models according to the Artificial Analysis Intelligence Index. The key advantage of o1 Pro lies in its increased computational power, which delivers more reliable and precise results. This capability could make it the optimal choice for certain use cases, as no comparable offering currently exists.

AI News Without the Hype – Curated by Humans

As a THE DECODER subscriber, you get ad-free reading, our weekly AI newsletter, the exclusive "AI Radar" Frontier Report 6× per year, access to comments, and our complete archive.