On September 12, 2024, OpenAI’s newest models “o1-preview” and “o1-mini” were introduced. Soon, these models will join Microsoft Azure OpenAI Service, Azure AI Studio, and GitHub Models.
What are OpenAI o1-preview and o1-mini models?
The OpenAI o1-preview and o1-mini models are specifically designed to tackle reasoning and problem-solving tasks with increased focus and capability. These models spend more time processing and understanding the user’s request. This makes them exceptionally strong in areas like science, coding, and math.
- o1-preview: Designed for advanced reasoning and solving complex problems, including math and science tasks. Ideal for applications requiring deep contextual understanding and agentic workflows.
- o1-mini: A smaller, faster, and 80% cheaper model compared to o1-preview. It excels in code generation and small context operations.
Key capabilities of the OpenAI o1 series
- Complex Code Generation: Capable of generating algorithms and handling advanced coding tasks to support developers.
- Advanced Problem Solving: Ideal for comprehensive brainstorming sessions and addressing multifaceted challenges.
- Complex Document Comparison: Perfect for analyzing contracts, case files, or legal documents to identify subtle differences.
- Instruction Following and Workflow Management: Particularly effective for managing workflows requiring shorter contexts.
Use Cases
For example, o1 can be used by healthcare researchers to annotate cell sequencing data. It can help physicists to generate complicated mathematical formulas needed for quantum optics. It is also interesting for developers wanting to build and execute multi-step workflows.
Current Limitations
o1-preview and o1-mini do not allow web browsing, file uploads or structured outputs which are available in the latest GPT-4o model. These features will follow soon for ChatGPT and Azure. For many common tasks, GPT-4o remains more capable in the near term. However, for complex reasoning tasks, the o1 series represents a significant advancement.
Comparing GPT-4o, o1-mini and o1-preview
The new o1 models excel in challenging tasks. The chart below compares the test results of GPT-4o, o1-mini, and o1-preview across various coding and Science, Technology, Engineering, and Mathematics (STEM) tests. A wider bar indicates better performance compared to the other models.
Availability via Azure
We expect them to become generally available in the coming weeks in Azure.
Security
o1-preview and o1-mini models in Azure OpenAI Service have Content Safety features activated by default, at no additional cost for customers.
FAQ
What are the main differences between o1-preview and o1-mini?
o1-preview is focused on advanced reasoning and solving complex problems, while o1-mini is smaller, faster, and more cost-effective and excels in code generation and small context operations.
What are the key applications of these models?
The o1 series is ideal for complex coding, math reasoning, brainstorming, and comparative analysis. They are particularly useful for advanced problem-solving and document comparison.
Is there any way to try them already?
Yes. OpenAI is planning to bring o1-mini access to all ChatGPT Free users so you can try it out in ChatGPT (without data protection!).
More information
The announcement can be found here: https://azure.microsoft.com/en-us/blog/introducing-o1-openais-new-reasoning-model-series-for-developers-and-enterprises-on-azure/.
Access o1-preview on Azure: https://ai.azure.com/explore/models/o1-preview/version/1/registry/azure-openai.
Access o1-mini on Azure: https://ai.azure.com/explore/models/o1-mini/version/1/registry/azure-openai.
Find an article on reasoning with OpenAI o1 here: https://openai.com/index/learning-to-reason-with-llms/.
To learn about Microsoft Licensing with SCHNEIDER IT MANAGEMENT, please visit: https://www.schneider.im/software/microsoft.
Please contact us for expert services on your specific Microsoft software and online services requirements and to request a quote today.