OpenAI offers DALL-E in free tier, but with daily limit on image generation

OpenAI launched the DALL-E 3 generative AI model in September last year. However, access to the third generation image generation model was limited to subscribers in paid tiers

OpenAI is now allowing free-tier users to access image generation AI model DALL-E 3. In a post on social media platform X, the Microsoft-backed US-based artificial intelligence company stated that it is rolling out the ability to generate up to two images per day for ChatGPT free users through the DALL-E 3 model. Free users can prompt the ChatGPT chatbot to create an image based on their input.

OpenAI launched the DALL-E 3 generative AI model in September last year. However, access to the third-generation image generation model was limited to ChatGPT Plus subscribers. One of the key features of the DALL-E 3 model is that it allows users to make accurate prompts for generating desired images through ChatGPT’s conversational interface. Text generating ability of the AI-chatbot removes some of the complexity and refines the input prompt that the model then uses to generate images.

Earlier this year, OpenAI announced that users can also edit sections of DALL-E 3 generated images. The company introduced a new selection tool, which can be accessed by clicking on a generated image. With this new tool, users can select an area in the image to edit and then describe the changes in chat in the form of a text prompt.

In related news, OpenAI has released “System Card” for its newest multimodal AI model GPT-4o. This document highlights the safety measures taken by the company and evaluation report of the AI model. According to the document, OpenAI evaluated risks such as unauthorised voice generation, generating violent speech, generation of copyright music and more. It stated that the process of evaluation resulted in the GPT-4o model receiving an overall “Medium Risk” rating. This conclusion was drawn from rating among four overall categories – cybersecurity, biological threats, model autonomy and persuasion. While OpenAI’s latest model was rated low risk in the first three categories, research showed that GPT-4o generated content could be persuasive.