Image: OpenAI.
If you’ve ever used a text-to-image generator, you’re probably familiar with DALL-E, the model from OpenAI that sparked interest in image generators in the first place. Last month, OpenAI announced an even more advanced version of DALL-E, available now.
On Monday, OpenAI launched a beta version of DALL-E 3, as can be read in the release notes. DALL-E 3 is therefore available directly from ChatGPT, on web and on mobile, for certain users.
If OpenAI’s notes do not specify who DALL-E 3 is aimed at, the company had specified in September, when announcing this model, that it would be deployed for ChatGPT Plus and its Enterprise clients in October.
A more precise model
To access DALL-E 3 in ChatGPT, you will need to choose DALL-E 3 from the GPT-4 option drop-down menu.
Once this choice is made, you can enter your text. The description of the image you want DALL-E to offer you can be a short sentence or a detailed paragraph. Once your query is launched, DALL-E 3 generates the image in a few seconds, directly in ChatGPT.
When DALL-E 3 was announced last month, OpenAI said the new model would further exceed the capabilities of DALL-E 2 by creating more accurate images and better capturing the nuances of the user-submitted message or the user.
Browse with Bing exits beta
If you want to test the difference for yourself, DALL-E 3 is already available for free in Bing Chat. ZDNet also experimented with DALL-E 3 in Bing Chat. The result: rendered images are of higher quality, with greater attention to detail.
The release notes also reveal that the search feature in Bing, which was relaunched at the end of September after being removed from the platform due to user abuse, is coming out of its beta phase.
So, Plus and Enterprise users will be able to choose Browse with Bing in the GPT-4 model selector without having to switch to the beta version.
Source: ZDNet.com