OpenAI Presented Images 2.0 Model for Image Generation in ChatGPT
OpenAI has introduced Images 2.0, its first visual model featuring advanced 'reasoning' capabilities. The new model is now available to all users through the ChatGPT chatbot and via the API.
Key Features of Images 2.0
- Reasoning Mode: When enabled, the model can search the web for relevant information, generate multiple images from a single prompt, and verify its own results.
- Superior Language Support: Unlike previous models, Images 2.0 excels at rendering non-Latin scripts like Japanese, Korean, Chinese, Hindi, and Bengali, ensuring both accuracy and coherence.
- Enhanced Realism and Style: The model adds subtle imperfections to photos for a more realistic look and accurately reproduces a wide range of styles, including cinematic shots and pixel art.
Availability and Limitations
While Images 2.0 is accessible to all ChatGPT and Codex users, the advanced reasoning mode is exclusive to ChatGPT Plus, Pro, and Business subscribers. The gpt-image-2 model is also accessible via the API.
OpenAI acknowledges that the model may still face challenges with highly complex tasks, such as generating origami instructions, puzzles, or rendering fine details on inclined surfaces.
Image Generation Showcase
