OpenAI launches $200/month ChatGPT Pro with 'thinking harder' mode

OpenAI just drew a bright line between casual users and the research crowd. ChatGPT Pro is a new $200 monthly tier, and the price tag isn’t just for unlimited access to existing models. It’s the entry fee for o1 pro mode, a version of OpenAI’s reasoning model that burns more compute to ‘think harder’ on the toughest problems.

The company says this mode isn’t about speed. It’s about reliability. External testers found o1 pro mode delivered more consistently accurate answers in data science, programming, and case law analysis. OpenAI used a stricter benchmark where the model had to nail a question four times out of four to count as a success. On competition math (AIME 2024), o1 pro mode hit 86% under that 4/4 reliability metric. The standard o1 managed 50%, and o1-preview scored 44%. For PhD-level science questions, the gap was similar but tighter: 79% for pro mode versus 74% and 73% for the others.

The interface signals how different this is. Answers take long enough that ChatGPT shows a progress bar and pushes an in-app notification if you switch conversations. That’s a clear message: this isn’t for quick queries. It’s batch processing for your hardest thinking.

OpenAI is seeding adoption in a targeted way — ten grants to medical researchers at places like Boston Children’s Hospital, Harvard Medical School, and Berkeley Lab. Catherine Brownstein, who hunts for rare disease genes, and Justin Reese, who mines biomedical data, are among the first recipients. It’s a smart move. Nothing validates a premium research tool like putting it in the hands of people who might actually cure something with it. The question now is whether the broader market of engineers and analysts will see enough ROI at $200 a month, especially when the main selling point is reliability rather than a flashy new capability.