Official

openai / o1

OpenAI's first o-series reasoning model

  • Public
  • 3.1K runs
  • License
Iterate in playground

Pricing

Official model
Pricing for official models works differently from other models. Instead of being billed by time, you’re billed by input and output, making pricing more predictable.

This model is priced by how many input tokens are sent and how many output tokens are generated.

Check out our docs for more information about how per-token pricing works on Replicate.

Readme

OpenAI o1 is the most powerful language model available in ChatGPT as of 2025, designed to solve the most complex and high-stakes problems in research, engineering, and beyond. Optimized for depth, accuracy, and reliability, o1 represents a new tier of research-grade intelligence — with o1 pro mode pushing the boundaries even further through higher compute and more deliberative reasoning.


Core Strengths

  • Research-grade intelligence with cutting-edge performance
  • Built for high-stakes tasks in law, medicine, data science, and engineering
  • o1 pro mode enables deeper, more reliable thinking with extra compute
  • Powers ChatGPT’s Pro experience for professionals and researchers
  • Outperforms previous models (GPT-4o, GPT-4-turbo) on hard problems
  • Evaluated using stricter reliability metrics (4/4 accuracy)

Model Performance

Competition Math (AIME 2024):         
- o1-preview:     50  
- o1:             78  
- o1 pro mode:    86  

Competition Code (Codeforces):        
- o1-preview:     62  
- o1:             89  
- o1 pro mode:    90  

PhD-Level Science (GPQA Diamond):     
- o1-preview:     74  
- o1:             76  
- o1 pro mode:    79  
Competition Math (AIME 2024):         
- o1-preview:     37  
- o1:             67  
- o1 pro mode:    80  

Competition Code (Codeforces):        
- o1-preview:     26  
- o1:             64  
- o1 pro mode:    75  

PhD-Level Science (GPQA Diamond):     
- o1-preview:     58  
- o1:             67  
- o1 pro mode:    74  

Use Cases

  • Research-grade analysis in data science, law, and medicine
  • Debugging and deep investigation in advanced software engineering
  • Complex reasoning in math, science, and multimodal tasks
  • Writing and validating technical content with 4/4 reliability
  • High-compute applications that benefit from extra deliberation time