o1-pro

Artificial Analysis flags o1-pro as deprecated, stating it only continues performance benchmarking for the default 10k input token workload and that results for other workloads are historical and no longer updated. The page recommends considering OpenAI's newer o3-pro model instead and links to comparison and API provi Beyond the deprecation notice, Artificial Analysis records o1-pro specifications as a proprietary OpenAI model released March 2025 with an Artificial Analysis Intelligence Index score of 12, priced at $150.00 per 1M input tokens and $600.00 per 1M output tokens, supporting text and image input and text output with a 20

A peer-reviewed study in Clinics benchmarked DeepSeek-R1 against three OpenAI models, including OpenAI o1 pro, on 321 text-based USMLE-style questions. DeepSeek-R1 achieved the highest overall accuracy at 92.5%, while the OpenAI models (including o1 pro) each scored 78.8%, with the gap statistically significant (p < 0. The authors conclude that DeepSeek-R1 surpasses both OpenAI models and reported average human examinee performance across all three USMLE steps, positioning it as a compelling tool for medical education. The study nonetheless notes strong inter-model consensus, with agreement between DeepSeek-R1 and OpenAI o1 pro reach

A peer-reviewed comparative benchmark published in Nature Scientific Reports evaluated o1 Pro against DeepSeek R1, DeepSeek-R1-Lite, and Grok 3 on 493 ophthalmology board-style questions drawn from StatPearls and EyeQuiz. o1 Pro achieved the highest overall accuracy at 83.4%, significantly outperforming DeepSeek-R1-Lit The authors framed the result as evidence that first-generation reasoning models like o1 Pro generalize from standard math and science benchmarks to high-complexity clinical knowledge, while noting that real-world performance, clinical integration, and human-AI interaction effects remain open questions for medical educ