Ticker

6/recent/ticker-posts

You will hallucinate in front of the new images that Chatgpt can create, on a small condition

You will hallucinate in front of the new images that Chatgpt can create, on a small condition

OpenAI has just announced a major update to ChatGPT. The popular chatbot is now capable of generating images at a level never before seen. But be careful, not everyone will be able to benefit from it right away.

You will hallucinate in front of the new images that Chatgpt can create, on a small condition

ChatGPT is constantly evolving and now offers much more than just text responses. After the recent release of GPT-4.5, it's the turn of GPT-4o to make headlines with a major advance: the generation of high-quality images. This new model is gradually replacing DALL-E 3, with much more detailed visuals, coherent scenes and the ability to finely modify the elements of an image. This new feature places the chatbot among the most powerful graphic creation tools.

Thanks to this update With the latest update, GPT-4o users can now create realistic images directly in their conversations, without changing templates or opening a new window. The tool also allows you to edit an image, modify its style, colors, or even isolate certain elements from the background. It becomes possible to place readable text with precision, add subtle effects, and maintain continuity between generated images. This is a significant improvement over DALL-E 3, whose text generation and fidelity details often left something to be desired.

GPT-4o allows for image generation with unprecedented quality

Despite the enthusiasm surrounding this update, not everyone can benefit from it yet. OpenAI confirms that only subscribers to ChatGPT Plus currently have access to image generation with GPT-4o. This is due to very high demand and limited resources. The option is expected to gradually be extended to free users, but no date has been announced. For now, you need to subscribe to a €20 per month subscription to test this new feature.

This update also marks a strategic change for OpenAI. DALL-E 3 officially becomes a "legacy" model, meaning destined to disappear. Eventually, GPT-4o will take over entirely for visual creation. To develop this new system, OpenAI relies on user feedback, but also on correction work carried out by around a hundred people to refine the results. An approach that bears fruit, both in terms of quality and quality. The generated images already impress. the first testers.

Post a Comment

0 Comments