OpenAI has introduced o3‑Pro, an upgraded version of its o3 reasoning model, now available to ChatGPT Pro and Team users, as well as via its developer API. Designed to deliver more reliable and precise reasoning across complex domains, o3‑Pro outperforms o3 in expert evaluations on science, education, programming, business, and writing tasks.
Priced at $20 per million input tokens and $80 per million output tokens, o3‑Pro includes expanded tool access—enabling web search, file analysis, visual reasoning, Python execution, and memory personalization. While it delivers significantly better accuracy, responses are slower compared to the o1‑Pro model.
OpenAI notes that some features—like temporary Chat sessions, image generation, and Canvas—are temporarily disabled for o3‑Pro. Additionally, it replaces the o1‑Pro model in the selector for Pro and Team tiers, with Enterprise and Education users gaining access the following week.
On the performance front, o3‑Pro delivers strong results across key AI benchmarks, based on OpenAI’s internal evaluations. It outperforms Google’s Gemini 2.5 Pro on AIME 2024, a benchmark for mathematical reasoning, and surpasses Anthropic’s Claude 4 Opus on GPQA Diamond, which assesses PhD-level science expertise.
Quick Take
OpenAI’s rollout of o3‑Pro solidifies its lead in advanced reasoning models, prioritizing accuracy and depth over speed. By enabling full tool integration, it shifts ChatGPT closer to autonomous agents capable of complex analysis. The premium pricing and slower response times, however, position o3‑Pro as a specialist model tailored for high-stakes professional use.
