Integrated Tool Use in Chain-of-Thought
Supports web browsing, Python code execution, image and file analysis, image generation, canvas, automations, file search, and memory within its reasoning process.
Visual Reasoning with Image Transformations
Processes images natively with cropping, zooming, and rotating capabilities during reasoning without separate models.
Adjustable Reasoning Effort and Debugging
Allows users to set reasoning effort levels (low, medium, high) and access full chain-of-thought for debugging purposes.
Large Context Window and Multi-Modal Input
Supports a 200,000 token context window and accepts both text and image inputs with text outputs.
Function Calling and Structured Outputs
Enables function calling and structured output formats to facilitate integration and automation.