OpenAI's new models can 'think with pictures'
Briefly

OpenAI has introduced two new AI models, o3 and 04-mini, designed to excel in programming, math, and science by incorporating advanced reasoning capabilities. These models can interpret low-quality images, allowing users to upload sketches or diagrams that the models will comprehend. Additionally, o3 and 04-mini are capable of generating images and browsing autonomously. To access these advanced features, users must have a Chat GPT Plus, Pro, or Team subscription, with plans for a more powerful o3-pro model set to be released soon.
OpenAI’s new o3 and 04-mini models enable enhanced reasoning in programming, math, and science, with the ability to interpret low-quality images.
Users can upload sketches or diagrams, and despite low quality, o3 and 04-mini can successfully interpret and understand the content.
These models can generate images and browse independently, adding significant versatility to their functionality.
Accessing o3 and 04-mini requires a Chat GPT Plus, Pro, or Team subscription, with a more powerful o3-pro model forthcoming.
Read at Computerworld
[
|
]