The American company OpenAI has announced a significant update to its ChatGPT system, introducing a new image generation capability based on the improved GPT-4o model, writes IZ. This feature allows you to create images with high accuracy that match the given prompts and the context of the chat. The company emphasizes that the new model will significantly improve the quality of the generated images compared to the previous version, which used the Dall-E 3 neural network.
The company said in a statement that GPT-4o makes image generation a more accurate and efficient process, allowing it to create images that accurately match user descriptions. The new version of the model generates images with up to 20 objects at a time, allowing for a higher level of detail and control over composition.
OpenAI also said that GPT-4o can integrate user-uploaded images into its context to further generate new variants. The company promised to continue blocking requests to generate images that violate its policy on objectionable content, such as deepfakes or sexually explicit material.
The model is available now to non-subscribers, as well as Enterprise and Edu subscribers. GPT-4o will also be made available via API in the coming weeks.
As a reminder, we previously wrote that the Apple Watch could get cameras and artificial intelligence.