OpenAI’s API users now have full access to the new o1 model

Core model improvements: OpenAI is granting API users full access to its latest o1 model, replacing the previous o1-preview version with enhanced capabilities and efficiency.

• The upgraded model restores key features including developer message guidance for chatbot customization and adjustable “reasoning effort” parameters
• Visual information processing capabilities now allow the model to analyze document scans
• The o1 model demonstrates 60% reduction in thinking tokens while achieving 25-35 percentage point improvements in accuracy on benchmark tests
• Tier 5 development customers will receive immediate access, while o1 Pro model API access remains pending

Technical enhancements and cost savings: The API upgrade introduces significant improvements to function calling and structured output handling while reducing operational costs.

• Internal function calling capabilities have been refined to better integrate with developer-created functions
• Structured outputs using JSON schema show increased accuracy in formatting
• Audio token costs have been reduced by 60% for o1 and 90% for 4o mini tokens
• New WebRTC support simplifies audio interface development, reducing code requirements from 250 to approximately 12 lines

Developer tools and accessibility: OpenAI has expanded its development ecosystem with new tools and programming language support.

• New Software Development Kits (SDKs) for Go and Java programming languages have been introduced
• Plug-and-play WebRTC code will be available for implementing AI assistants in various devices
• The platform now supports implementation in devices ranging from toy reindeer to smart glasses and context-aware cameras

Fine-tuning innovations: A new approach to model customization has been implemented through direct preference optimization.

• The new system allows developers to indicate preferred responses rather than providing exact input/output pairs
• The process automatically identifies and optimizes for differences in verbosity, formatting, style guidelines, and response quality
• This represents a significant shift from traditional supervised fine-tuning methods

Looking ahead: While these updates mark substantial progress in AI development tools, the pending release of o1 Pro model access and the simplified fine-tuning approach suggest OpenAI is strategically positioning itself to make AI development more accessible while maintaining high performance standards.